diff --git a/changelog_entry.yaml b/changelog_entry.yaml index e69de29..a90a995 100644 --- a/changelog_entry.yaml +++ b/changelog_entry.yaml @@ -0,0 +1,5 @@ +- bump: minor + changes: + added: + - Error bars and grid lines to visualizations. + - Notebook benchmarking models on additional datasets. diff --git a/docs/imputation-benchmarking/cross-validation.md b/docs/imputation-benchmarking/cross-validation.md index 134772d..5e440e8 100644 --- a/docs/imputation-benchmarking/cross-validation.md +++ b/docs/imputation-benchmarking/cross-validation.md @@ -44,20 +44,28 @@ Returns a dictionary containing separate results for each metric type: ```python { "quantile_loss": { - "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles + "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (mean across folds) + "results_std": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (std across folds) "mean_train": float, "mean_test": float, - "variables": List[str] # numerical variables evaluated + "std_train": float, + "std_test": float, + "variables": List[str] # numerical variables evaluated }, "log_loss": { - "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles + "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles + "results_std": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (std across folds) "mean_train": float, "mean_test": float, - "variables": List[str] # categorical variables evaluated + "std_train": float, + "std_test": float, + "variables": List[str] # categorical variables evaluated } } ``` +The `results_std` DataFrame and `std_train`/`std_test` values provide the standard deviation of the loss across cross-validation folds, which can be used to visualize uncertainty via error bars. + If `tune_hyperparameters=True`, returns a tuple of `(results_dict, best_hyperparameters)`. ## Example usage diff --git a/docs/imputation-benchmarking/visualizations.md b/docs/imputation-benchmarking/visualizations.md index 86867ac..d6c9450 100644 --- a/docs/imputation-benchmarking/visualizations.md +++ b/docs/imputation-benchmarking/visualizations.md @@ -37,6 +37,7 @@ class MethodComparisonResults: show_mean: bool = True, figsize: Tuple[int, int] = (PLOT_CONFIG["width"], PLOT_CONFIG["height"]), plot_type: str = "bar", + show_error_bars: bool = True, ) -> go.Figure def summary(self, format: str = "wide") -> pd.DataFrame @@ -53,6 +54,7 @@ class MethodComparisonResults: | show_mean | bool | True | Show horizontal lines for mean loss | | figsize | tuple | (width, height) | Figure dimensions in pixels | | plot_type | str | "bar" | Plot type: "bar" for grouped bars, "stacked" for contribution analysis | +| show_error_bars | bool | True | Show error bars representing standard deviation across CV folds | The `"stacked"` plot type shows rank-based contribution scores, useful for understanding how each variable contributes to overall model performance. @@ -134,6 +136,7 @@ class PerformanceResults: title: Optional[str] = None, save_path: Optional[str] = None, figsize: Tuple[int, int] = (PLOT_CONFIG["width"], PLOT_CONFIG["height"]), + show_error_bars: bool = True, ) -> go.Figure def summary(self) -> pd.DataFrame @@ -146,6 +149,7 @@ class PerformanceResults: | title | str | None | Custom plot title | | save_path | str | None | Path to save the plot | | figsize | tuple | (width, height) | Figure dimensions in pixels | +| show_error_bars | bool | True | Show error bars representing standard deviation across CV folds | For quantile loss, the plot shows train and test loss across quantiles as grouped bars. For log loss, the plot includes the loss bars and optionally confusion matrix and class probability distribution subplots. For combined metrics, both are shown in subplots. diff --git a/microimpute/__init__.py b/microimpute/__init__.py index 170ef61..2667c7e 100644 --- a/microimpute/__init__.py +++ b/microimpute/__init__.py @@ -72,6 +72,11 @@ except ImportError: pass +try: + from microimpute.models.mdn import MDN +except ImportError: + pass + # Import visualization modules from microimpute.visualizations import ( MethodComparisonResults, diff --git a/microimpute/config.py b/microimpute/config.py index bcd51b3..c2b0eb7 100644 --- a/microimpute/config.py +++ b/microimpute/config.py @@ -84,5 +84,28 @@ PLOT_CONFIG: Dict[str, Any] = { "width": 750, "height": 600, - "colors": {}, + # Plotly Safe palette - colorblind-friendly + "color_palette": [ + "#88CCEE", # Cyan + "#CC6677", # Rose + "#DDCC77", # Sand + "#117733", # Green + "#332288", # Indigo + "#AA4499", # Purple + "#44AA99", # Teal + "#999933", # Olive + "#882255", # Wine + "#661100", # Brown + ], + # Background colors (same for both) + "plot_bgcolor": "#FAFAFA", + "paper_bgcolor": "#FAFAFA", + # Grid styling (horizontal only) + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "showgrid_x": False, + "showgrid_y": True, + # Axis line styling + "linecolor": "#CCCCCC", + "showline": True, } diff --git a/microimpute/evaluations/cross_validation.py b/microimpute/evaluations/cross_validation.py index 371ca1d..664adb8 100644 --- a/microimpute/evaluations/cross_validation.py +++ b/microimpute/evaluations/cross_validation.py @@ -494,15 +494,21 @@ def cross_validate_model( Dictionary containing separate results for quantile_loss and log_loss: { "quantile_loss": { - "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles + "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (mean across folds) + "results_std": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (std across folds) "mean_train": float, "mean_test": float, + "std_train": float, # std of mean loss across folds + "std_test": float, # std of mean loss across folds "variables": List[str] }, "log_loss": { "results": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (constant values) + "results_std": pd.DataFrame, # rows: ["train", "test"], cols: quantiles (std across folds) "mean_train": float, "mean_test": float, + "std_train": float, + "std_test": float, "variables": List[str] } } @@ -676,26 +682,54 @@ def cross_validate_model( index=["train", "test"], ) - # Calculate means + # Create std DataFrame for error bars + std_df = pd.DataFrame( + [ + { + q: np.std(values) if len(values) > 1 else 0.0 + for q, values in metric_results[metric_type][ + "train" + ].items() + }, + { + q: np.std(values) if len(values) > 1 else 0.0 + for q, values in metric_results[metric_type][ + "test" + ].items() + }, + ], + index=["train", "test"], + ) + + # Calculate means and stds across all quantiles mean_test = combined_df.loc["test"].mean() mean_train = combined_df.loc["train"].mean() + std_test = std_df.loc["test"].mean() + std_train = std_df.loc["train"].mean() final_results[metric_type] = { "results": combined_df, # Single DataFrame with train/test rows + "results_std": std_df, # Std across folds for each quantile "mean_train": mean_train, "mean_test": mean_test, + "std_train": std_train, + "std_test": std_test, "variables": metric_results[metric_type]["variables"], } log.info( - f"{metric_type} - Mean Train: {mean_train:.6f}, Mean Test: {mean_test:.6f}" + f"{metric_type} - Mean Train: {mean_train:.6f} (±{std_train:.6f}), " + f"Mean Test: {mean_test:.6f} (±{std_test:.6f})" ) else: # No variables use this metric final_results[metric_type] = { "results": pd.DataFrame(), # Empty DataFrame + "results_std": pd.DataFrame(), # Empty DataFrame "mean_train": np.nan, "mean_test": np.nan, + "std_train": np.nan, + "std_test": np.nan, "variables": [], } diff --git a/microimpute/models/mdn.py b/microimpute/models/mdn.py index 7558e7f..096e18c 100644 --- a/microimpute/models/mdn.py +++ b/microimpute/models/mdn.py @@ -291,6 +291,9 @@ def fit( verbose=False, suppress_lightning_logger=True, ) + # Disable Lightning's default CSVLogger to avoid + # "dict contains fields not in fieldnames" errors + self.model.logger = False self.model.fit(train=train_data) @@ -477,6 +480,9 @@ def fit( verbose=False, suppress_lightning_logger=True, ) + # Disable Lightning's default CSVLogger to avoid + # "dict contains fields not in fieldnames" errors + self.model.logger = False self.model.fit(train=train_data) diff --git a/microimpute/models/quantreg.py b/microimpute/models/quantreg.py index 1d701e9..2f2c686 100644 --- a/microimpute/models/quantreg.py +++ b/microimpute/models/quantreg.py @@ -32,6 +32,7 @@ def __init__( quantiles_specified: bool = False, boolean_targets: Optional[Dict[str, Dict]] = None, constant_targets: Optional[Dict[str, Dict]] = None, + dummy_processor: Optional[Any] = None, ) -> None: """Initialize the QuantReg results. @@ -46,6 +47,7 @@ def __init__( names before dummy encoding. quantiles_specified: Whether quantiles were explicitly specified during fit. boolean_targets: Dictionary of boolean target info for conversion back to bool. + dummy_processor: Processor for handling dummy encoding in test data. """ super().__init__( predictors, @@ -59,6 +61,7 @@ def __init__( self.quantiles_specified = quantiles_specified self.boolean_targets = boolean_targets or {} self.constant_targets = constant_targets or {} + self.dummy_processor = dummy_processor @validate_call(config=VALIDATE_CONFIG) def _predict( @@ -414,6 +417,7 @@ def _fit( quantiles_specified=(quantiles is not None), boolean_targets=boolean_targets, constant_targets=constant_targets, + dummy_processor=getattr(self, "dummy_processor", None), ) except Exception as e: self.logger.error(f"Error fitting QuantReg model: {str(e)}") diff --git a/microimpute/utils/dashboard_formatter.py b/microimpute/utils/dashboard_formatter.py index c8847d4..fe8dbb6 100644 --- a/microimpute/utils/dashboard_formatter.py +++ b/microimpute/utils/dashboard_formatter.py @@ -179,6 +179,7 @@ def _format_histogram_rows( "quantile": "N/A", "metric_name": "categorical_distribution", "metric_value": None, # Not used for histograms + "metric_std": None, "split": "full", "additional_info": json.dumps( { @@ -211,6 +212,7 @@ def _format_histogram_rows( "quantile": "N/A", "metric_name": "histogram_distribution", "metric_value": None, # Not used for histograms + "metric_std": None, "split": "full", "additional_info": json.dumps( { @@ -375,6 +377,7 @@ def format_csv( if metric_type in cv_result: data = cv_result[metric_type] results_df = data.get("results") + results_std_df = data.get("results_std") variables = data.get("variables", []) if results_df is not None: @@ -382,7 +385,18 @@ def format_csv( for split in ["train", "test"]: if split in results_df.index: for quantile in results_df.columns: - # Add mean_all row for this quantile + # Get std value if available + std_value = None + if ( + results_std_df is not None + and split in results_std_df.index + ): + std_value = float( + results_std_df.loc[ + split, quantile + ] + ) + rows.append( { "type": "benchmark_loss", @@ -393,6 +407,7 @@ def format_csv( "metric_value": results_df.loc[ split, quantile ], + "metric_std": std_value, "split": split, "additional_info": json.dumps( { @@ -406,6 +421,11 @@ def format_csv( # Add mean across all quantiles if "mean_train" in data: + std_train = ( + float(data["std_train"]) + if "std_train" in data + else None + ) rows.append( { "type": "benchmark_loss", @@ -414,6 +434,7 @@ def format_csv( "quantile": "mean", "metric_name": metric_type, "metric_value": data["mean_train"], + "metric_std": std_train, "split": "train", "additional_info": json.dumps( {"n_variables": len(variables)} @@ -422,6 +443,11 @@ def format_csv( ) if "mean_test" in data: + std_test = ( + float(data["std_test"]) + if "std_test" in data + else None + ) rows.append( { "type": "benchmark_loss", @@ -430,6 +456,7 @@ def format_csv( "quantile": "mean", "metric_name": metric_type, "metric_value": data["mean_test"], + "metric_std": std_test, "split": "test", "additional_info": json.dumps( {"n_variables": len(variables)} @@ -461,6 +488,13 @@ def format_csv( else: quantile = float(percentile) if pd.notna(percentile) else "N/A" + # Get std if available in the DataFrame + std_value = ( + float(row["Std"]) + if "Std" in row.index and pd.notna(row["Std"]) + else None + ) + rows.append( { "type": "benchmark_loss", @@ -469,6 +503,7 @@ def format_csv( "quantile": quantile, "metric_name": row["Metric"], "metric_value": row["Loss"], + "metric_std": std_value, "split": "test", # Comparison metrics are typically on test set "additional_info": "{}", } @@ -490,6 +525,7 @@ def format_csv( .lower() .replace(" ", "_"), # e.g., "wasserstein_distance" "metric_value": row["Distance"], + "metric_std": None, "split": "full", "additional_info": "{}", } @@ -514,6 +550,7 @@ def format_csv( "quantile": "N/A", "metric_name": corr_type, "metric_value": corr_matrix.iloc[i, j], + "metric_std": None, "split": "full", "additional_info": json.dumps( {"predictor2": pred2} @@ -534,6 +571,7 @@ def format_csv( "quantile": "N/A", "metric_name": "mutual_info", "metric_value": mi_df.loc[predictor, target], + "metric_std": None, "split": "full", "additional_info": json.dumps({"target": target}), } @@ -556,6 +594,7 @@ def format_csv( "quantile": "N/A", "metric_name": "relative_impact", "metric_value": row["relative_impact"], + "metric_std": None, "split": "test", "additional_info": json.dumps( {"removed_predictor": predictor} @@ -572,6 +611,7 @@ def format_csv( "quantile": "N/A", "metric_name": "loss_increase", "metric_value": row["loss_increase"], + "metric_std": None, "split": "test", "additional_info": json.dumps( {"removed_predictor": predictor} @@ -603,6 +643,7 @@ def format_csv( "quantile": "N/A", "metric_name": "cumulative_improvement", "metric_value": row["cumulative_improvement"], + "metric_std": None, "split": "test", "additional_info": json.dumps( { @@ -632,6 +673,7 @@ def format_csv( "quantile": "N/A", "metric_name": "marginal_improvement", "metric_value": row["marginal_improvement"], + "metric_std": None, "split": "test", "additional_info": json.dumps( { @@ -693,6 +735,7 @@ def format_csv( "quantile", "metric_name", "metric_value", + "metric_std", "split", "additional_info", ] diff --git a/microimpute/visualizations/comparison_plots.py b/microimpute/visualizations/comparison_plots.py index 46dbeef..4e1b85a 100644 --- a/microimpute/visualizations/comparison_plots.py +++ b/microimpute/visualizations/comparison_plots.py @@ -192,17 +192,29 @@ def _process_dual_metrics_input( # Get test results (single row) if "test" in ql_data["results"].index: test_results = ql_data["results"].loc["test"] + # Get std results if available + std_results = None + if ( + ql_data.get("results_std") is not None + and "test" in ql_data["results_std"].index + ): + std_results = ql_data["results_std"].loc["test"] + for quantile in test_results.index: for var in ql_data.get("variables", ["y"]): - long_format_data.append( - { - "Method": method_name, - "Imputed Variable": var, - "Percentile": quantile, - "Loss": test_results[quantile], - "Metric": "quantile_loss", - } - ) + row = { + "Method": method_name, + "Imputed Variable": var, + "Percentile": quantile, + "Loss": test_results[quantile], + "Metric": "quantile_loss", + "Std": ( + std_results[quantile] + if std_results is not None + else np.nan + ), + } + long_format_data.append(row) # Add mean loss if "mean_test" in ql_data: @@ -214,6 +226,7 @@ def _process_dual_metrics_input( "Percentile": "mean_quantile_loss", "Loss": ql_data["mean_test"], "Metric": "quantile_loss", + "Std": ql_data.get("std_test", np.nan), } ) @@ -230,6 +243,7 @@ def _process_dual_metrics_input( # Log loss is constant across quantiles if "test" in ll_data["results"].index: test_loss = ll_data["results"].loc["test"].mean() + test_std = ll_data.get("std_test", np.nan) for var in ll_data.get("variables", []): long_format_data.append( { @@ -238,6 +252,7 @@ def _process_dual_metrics_input( "Percentile": "log_loss", "Loss": test_loss, "Metric": "log_loss", + "Std": test_std, } ) @@ -251,6 +266,7 @@ def _process_dual_metrics_input( "Percentile": "mean_log_loss", "Loss": ll_data["mean_test"], "Metric": "log_loss", + "Std": ll_data.get("std_test", np.nan), } ) @@ -266,6 +282,7 @@ def plot( PLOT_CONFIG["height"], ), plot_type: str = "bar", + show_error_bars: bool = True, ) -> go.Figure: """Plot a comparison of performance across different imputation methods. @@ -275,6 +292,7 @@ def plot( show_mean: Whether to show horizontal lines for mean loss values. figsize: Figure size as (width, height) in pixels. plot_type: Type of plot: 'bar' (default) or 'stacked' (for contribution analysis) + show_error_bars: Whether to show error bars for variation across quantiles. Returns: Plotly figure object @@ -290,12 +308,14 @@ def plot( if plot_type == "stacked": return self._plot_stacked_contribution(title, save_path, figsize) elif self.metric == "log_loss": - return self._plot_log_loss_comparison(title, save_path, figsize) + return self._plot_log_loss_comparison( + title, save_path, figsize, show_error_bars + ) elif self.metric == "combined": return self._plot_combined_metrics(title, save_path, figsize) else: return self._plot_quantile_loss_comparison( - title, save_path, show_mean, figsize + title, save_path, show_mean, figsize, show_error_bars ) def _plot_quantile_loss_comparison( @@ -304,6 +324,7 @@ def _plot_quantile_loss_comparison( save_path: Optional[str], show_mean: bool, figsize: Tuple[int, int], + show_error_bars: bool = True, ) -> go.Figure: """Plot quantile loss comparison across methods.""" @@ -354,6 +375,12 @@ def _plot_quantile_loss_comparison( if title is None: title = f"Test {self.metric_name} Across Quantiles for Different Imputation Methods" + # Use Std column for error bars if requested + error_y_col = None + if show_error_bars and "Std" in melted_df.columns: + melted_df["Std"] = melted_df["Std"].fillna(0) + error_y_col = "Std" + # Create the bar chart logger.debug("Creating bar chart with plotly express") fig = px.bar( @@ -361,13 +388,14 @@ def _plot_quantile_loss_comparison( x="Percentile", y=self.metric_name, color="Method", - color_discrete_sequence=px.colors.qualitative.Plotly, + color_discrete_sequence=PLOT_CONFIG["color_palette"], barmode="group", title=title, labels={ "Percentile": "Quantiles", - self.metric_name: f"Test {self.metric_name}", + self.metric_name: f"{self.metric_name}", }, + error_y=error_y_col, ) # Add horizontal lines for mean loss if requested @@ -385,8 +413,8 @@ def _plot_quantile_loss_comparison( x1=n_percentiles - 0.5, y1=mean_loss, line=dict( - color=px.colors.qualitative.Plotly[ - i % len(px.colors.qualitative.Plotly) + color=PLOT_CONFIG["color_palette"][ + i % len(PLOT_CONFIG["color_palette"]) ], width=2, dash="dot", @@ -398,15 +426,29 @@ def _plot_quantile_loss_comparison( title_font_size=14, xaxis_title_font_size=12, yaxis_title_font_size=12, - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], legend_title="Method", height=figsize[1], width=figsize[0], ) - fig.update_xaxes(showgrid=False, zeroline=False) - fig.update_yaxes(showgrid=False, zeroline=False) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) # Save or show the plot if save_path: @@ -426,6 +468,7 @@ def _plot_log_loss_comparison( title: Optional[str], save_path: Optional[str], figsize: Tuple[int, int], + show_error_bars: bool = True, ) -> go.Figure: """Plot log loss comparison across methods.""" try: @@ -439,38 +482,55 @@ def _plot_log_loss_comparison( logger.warning("No log loss data available") return go.Figure() - # Get mean log loss per method - method_means = ( - log_loss_df.groupby("Method")["Loss"].mean().reset_index() + # Get mean log loss per method (Loss already contains the mean) + method_stats = log_loss_df.groupby("Method", as_index=False).agg( + {"Loss": "mean", "Std": "mean"} ) + method_stats["Std"] = method_stats["Std"].fillna(0) if title is None: - title = f"Log Loss Comparison Across Methods" + title = "Log Loss Comparison Across Methods" # Create bar chart + error_y_col = "Std" if show_error_bars else None fig = px.bar( - method_means, + method_stats, x="Method", y="Loss", color="Method", title=title, labels={"Loss": "Log Loss"}, - color_discrete_sequence=px.colors.qualitative.Plotly, + color_discrete_sequence=PLOT_CONFIG["color_palette"], + error_y=error_y_col, ) fig.update_layout( title_font_size=14, xaxis_title_font_size=12, yaxis_title_font_size=12, - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], height=figsize[1], width=figsize[0], showlegend=False, ) - fig.update_xaxes(showgrid=False, zeroline=False) - fig.update_yaxes(showgrid=False, zeroline=False) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) if save_path: _save_figure(fig, save_path) @@ -543,8 +603,8 @@ def _plot_combined_metrics( y=method_data["Loss"], name=method, legendgroup=method, - marker_color=px.colors.qualitative.Plotly[ - i % len(px.colors.qualitative.Plotly) + marker_color=PLOT_CONFIG["color_palette"][ + i % len(PLOT_CONFIG["color_palette"]) ], ), row=1, @@ -564,8 +624,8 @@ def _plot_combined_metrics( x=list(method_means.index), y=list(method_means.values), marker_color=[ - px.colors.qualitative.Plotly[ - i % len(px.colors.qualitative.Plotly) + PLOT_CONFIG["color_palette"][ + i % len(PLOT_CONFIG["color_palette"]) ] for i in range(len(method_means)) ], @@ -583,18 +643,46 @@ def _plot_combined_metrics( barmode="group", height=figsize[1] * 1.5, width=figsize[0], - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], showlegend=True, ) fig.update_xaxes( - title_text="Quantile", row=1, col=1, showgrid=False + title_text="Quantile", + row=1, + col=1, + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + fig.update_xaxes( + title_text="Method", + row=2, + col=1, + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], ) - fig.update_xaxes(title_text="Method", row=2, col=1, showgrid=False) - fig.update_yaxes(title_text="Loss", row=1, col=1, showgrid=False) fig.update_yaxes( - title_text="Log loss", row=2, col=1, showgrid=False + title_text="Loss", + row=1, + col=1, + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + fig.update_yaxes( + title_text="Log loss", + row=2, + col=1, + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], ) if save_path: @@ -726,13 +814,27 @@ def _plot_stacked_contribution( yaxis_title="Total rank score", height=figsize[1], width=figsize[0], - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], legend_title="Variable (Metric)", ) - fig.update_xaxes(showgrid=False, zeroline=False) - fig.update_yaxes(showgrid=False, zeroline=False) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) if save_path: _save_figure(fig, save_path) @@ -819,7 +921,6 @@ def __repr__(self) -> str: def method_comparison_results( data: Union[pd.DataFrame, Dict[str, Dict[str, Dict]]], - metric_name: Optional[str] = None, metric: str = "quantile_loss", data_format: str = "wide", ) -> MethodComparisonResults: diff --git a/microimpute/visualizations/performance_plots.py b/microimpute/visualizations/performance_plots.py index eda2a8f..8e9de4a 100644 --- a/microimpute/visualizations/performance_plots.py +++ b/microimpute/visualizations/performance_plots.py @@ -98,6 +98,7 @@ def plot( PLOT_CONFIG["width"], PLOT_CONFIG["height"], ), + show_error_bars: bool = True, ) -> go.Figure: """Plot the performance based on the specified metric. @@ -105,6 +106,7 @@ def plot( title: Custom title for the plot. If None, a default title is used. save_path: Path to save the plot. If None, the plot is displayed. figsize: Figure size as (width, height) in pixels. + show_error_bars: Whether to show error bars for variation across CV folds. Returns: Plotly figure object @@ -115,11 +117,17 @@ def plot( logger.debug(f"Creating performance plot for metric: {self.metric}") if self.metric == "quantile_loss": - return self._plot_quantile_loss(title, save_path, figsize) + return self._plot_quantile_loss( + title, save_path, figsize, show_error_bars + ) elif self.metric == "log_loss": - return self._plot_log_loss(title, save_path, figsize) + return self._plot_log_loss( + title, save_path, figsize, show_error_bars + ) elif self.metric == "combined": - return self._plot_combined(title, save_path, figsize) + return self._plot_combined( + title, save_path, figsize, show_error_bars + ) else: raise ValueError(f"Invalid metric: {self.metric}") @@ -128,13 +136,14 @@ def _plot_quantile_loss( title: Optional[str], save_path: Optional[str], figsize: Tuple[int, int], + show_error_bars: bool = True, ) -> go.Figure: """Plot quantile loss performance across quantiles.""" if not self.has_quantile_loss: logger.warning("No quantile loss data available") return go.Figure() - palette = px.colors.qualitative.Plotly + palette = PLOT_CONFIG["color_palette"] train_color = palette[2] test_color = palette[3] @@ -144,25 +153,48 @@ def _plot_quantile_loss( # Get the DataFrame for quantile loss ql_data = self.results["quantile_loss"]["results"] + # Get std data for error bars if available + ql_std = self.results["quantile_loss"].get("results_std") + # Add bars for training data if "train" in ql_data.index: + error_y_dict = None + if ( + show_error_bars + and ql_std is not None + and "train" in ql_std.index + ): + error_y_dict = dict( + type="data", array=ql_std.loc["train"].values + ) fig.add_trace( go.Bar( x=[str(x) for x in ql_data.columns], y=ql_data.loc["train"].values, name="Train", marker_color=train_color, + error_y=error_y_dict, ) ) # Add bars for test data if "test" in ql_data.index: + error_y_dict = None + if ( + show_error_bars + and ql_std is not None + and "test" in ql_std.index + ): + error_y_dict = dict( + type="data", array=ql_std.loc["test"].values + ) fig.add_trace( go.Bar( x=[str(x) for x in ql_data.columns], y=ql_data.loc["test"].values, name="Test", marker_color=test_color, + error_y=error_y_dict, ) ) @@ -176,14 +208,28 @@ def _plot_quantile_loss( barmode="group", width=figsize[0], height=figsize[1], - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], legend=dict(yanchor="top", y=0.99, xanchor="right", x=0.99), margin=dict(l=50, r=50, t=80, b=50), ) - fig.update_xaxes(showgrid=False, zeroline=False) - fig.update_yaxes(showgrid=False, zeroline=False) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) if save_path: _save_figure(fig, save_path) @@ -200,6 +246,7 @@ def _plot_log_loss( title: Optional[str], save_path: Optional[str], figsize: Tuple[int, int], + show_error_bars: bool = True, ) -> go.Figure: """Plot log loss performance and additional categorical metrics.""" if not self.has_log_loss: @@ -235,16 +282,23 @@ def _plot_log_loss( ) # Plot 1: Log Loss bars - palette = px.colors.qualitative.Plotly + palette = PLOT_CONFIG["color_palette"] train_color = palette[2] test_color = palette[3] # Get log loss values from the results DataFrame ll_results_df = ll_data["results"] + # Get std values for error bars + ll_std = ll_data.get("std_train") + ll_std_test = ll_data.get("std_test") + if "train" in ll_results_df.index: # Log loss should be constant across quantiles, so take the mean train_loss = ll_results_df.loc["train"].mean() + error_y_dict = None + if show_error_bars and ll_std is not None: + error_y_dict = dict(type="data", array=[ll_std]) fig.add_trace( go.Bar( x=["Train"], @@ -252,6 +306,7 @@ def _plot_log_loss( name="Train", marker_color=train_color, showlegend=True, + error_y=error_y_dict, ), row=1, col=1, @@ -259,6 +314,9 @@ def _plot_log_loss( if "test" in ll_results_df.index: test_loss = ll_results_df.loc["test"].mean() + error_y_dict = None + if show_error_bars and ll_std_test is not None: + error_y_dict = dict(type="data", array=[ll_std_test]) fig.add_trace( go.Bar( x=["Test"], @@ -266,6 +324,7 @@ def _plot_log_loss( name="Test", marker_color=test_color, showlegend=True, + error_y=error_y_dict, ), row=1, col=1, @@ -336,11 +395,26 @@ def _plot_log_loss( title=title, height=figsize[1] * num_subplots * 0.7, width=figsize[0], - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], showlegend=True, ) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + if save_path: _save_figure(fig, save_path) @@ -351,6 +425,7 @@ def _plot_combined( title: Optional[str], save_path: Optional[str], figsize: Tuple[int, int], + show_error_bars: bool = True, ) -> go.Figure: """Plot combined view of both metrics.""" if not self.has_quantile_loss and not self.has_log_loss: @@ -375,7 +450,7 @@ def _plot_combined( vertical_spacing=0.2, ) - palette = px.colors.qualitative.Plotly + palette = PLOT_CONFIG["color_palette"] train_color = palette[2] test_color = palette[3] current_row = 1 @@ -383,8 +458,18 @@ def _plot_combined( # Add quantile loss plot if self.has_quantile_loss: ql_data = self.results["quantile_loss"]["results"] + ql_std = self.results["quantile_loss"].get("results_std") if "train" in ql_data.index: + error_y_dict = None + if ( + show_error_bars + and ql_std is not None + and "train" in ql_std.index + ): + error_y_dict = dict( + type="data", array=ql_std.loc["train"].values + ) fig.add_trace( go.Bar( x=[str(x) for x in ql_data.columns], @@ -392,12 +477,22 @@ def _plot_combined( name="QL Train", marker_color=train_color, legendgroup="train", + error_y=error_y_dict, ), row=current_row, col=1, ) if "test" in ql_data.index: + error_y_dict = None + if ( + show_error_bars + and ql_std is not None + and "test" in ql_std.index + ): + error_y_dict = dict( + type="data", array=ql_std.loc["test"].values + ) fig.add_trace( go.Bar( x=[str(x) for x in ql_data.columns], @@ -405,6 +500,7 @@ def _plot_combined( name="QL Test", marker_color=test_color, legendgroup="test", + error_y=error_y_dict, ), row=current_row, col=1, @@ -417,9 +513,14 @@ def _plot_combined( # Add log loss plot if self.has_log_loss: ll_data = self.results["log_loss"]["results"] + ll_std_train = self.results["log_loss"].get("std_train") + ll_std_test = self.results["log_loss"].get("std_test") if "train" in ll_data.index: train_loss = ll_data.loc["train"].mean() + error_y_dict = None + if show_error_bars and ll_std_train is not None: + error_y_dict = dict(type="data", array=[ll_std_train]) fig.add_trace( go.Bar( x=["Log loss"], @@ -427,7 +528,8 @@ def _plot_combined( name="Log loss train", marker_color=train_color, legendgroup="train", - showlegend=self.has_quantile_loss == False, + showlegend=self.has_quantile_loss is False, + error_y=error_y_dict, ), row=current_row, col=1, @@ -435,6 +537,9 @@ def _plot_combined( if "test" in ll_data.index: test_loss = ll_data.loc["test"].mean() + error_y_dict = None + if show_error_bars and ll_std_test is not None: + error_y_dict = dict(type="data", array=[ll_std_test]) fig.add_trace( go.Bar( x=["Log loss"], @@ -442,7 +547,8 @@ def _plot_combined( name="Log loss test", marker_color=test_color, legendgroup="test", - showlegend=self.has_quantile_loss == False, + showlegend=self.has_quantile_loss is False, + error_y=error_y_dict, ), row=current_row, col=1, @@ -458,13 +564,27 @@ def _plot_combined( barmode="group", height=figsize[1] * num_subplots * 0.6, width=figsize[0], - paper_bgcolor="#F0F0F0", - plot_bgcolor="#F0F0F0", + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], showlegend=True, ) - fig.update_xaxes(showgrid=False, zeroline=False) - fig.update_yaxes(showgrid=False, zeroline=False) + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + gridwidth=PLOT_CONFIG["gridwidth"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + zeroline=False, + ) if save_path: _save_figure(fig, save_path) diff --git a/paper/benchmarking_datasets/benchmark_utils.py b/paper/benchmarking_datasets/benchmark_utils.py new file mode 100644 index 0000000..eb4489b --- /dev/null +++ b/paper/benchmarking_datasets/benchmark_utils.py @@ -0,0 +1,811 @@ +"""Benchmark utilities for CIA sensitivity analysis and cross-dataset comparison. + +This module provides functions for: +1. CIA (Conditional Independence Assumption) sensitivity analysis via progressive + predictor exclusion +2. Cross-dataset results visualization (summary tables and heatmaps) +""" + +import logging +from typing import Any, Dict, List, Optional, Type, Union + +import numpy as np +import pandas as pd +import plotly.graph_objects as go +from scipy.integrate import trapezoid +from sklearn.feature_selection import mutual_info_regression +from sklearn.model_selection import train_test_split +from tqdm.auto import tqdm + +from microimpute.comparisons.metrics import quantile_loss +from microimpute.config import ( + PLOT_CONFIG, + RANDOM_STATE, + TRAIN_SIZE, +) +from microimpute.models import Imputer + +log = logging.getLogger(__name__) + +# Method colors from PLOT_CONFIG (Safe palette - colorblind-friendly) +METHOD_COLORS = { + "QRF": "#88CCEE", # Cyan + "OLS": "#CC6677", # Rose + "QuantReg": "#DDCC77", # Sand + "Matching": "#117733", # Green + "MDN": "#332288", # Indigo +} + +# Use a simple set of quantiles for CIA analysis to avoid QuantReg issues +CIA_QUANTILES = [0.1, 0.3, 0.5, 0.7, 0.9] + + +def progressive_predictor_exclusion( + data: pd.DataFrame, + predictors: List[str], + imputed_variables: List[str], + model_class: Type[Imputer], + ordering: str = "mutual_info", + weight_col: Optional[Union[str, np.ndarray, pd.Series]] = None, + quantiles: Optional[List[float]] = None, + train_size: float = TRAIN_SIZE, + random_state: int = RANDOM_STATE, +) -> Dict[str, Any]: + """Progressively REMOVE predictors to measure CIA sensitivity. + + This function assesses sensitivity to the Conditional Independence Assumption + by measuring how performance degrades when key predictors are unavailable + (simulating incomplete linking variables between surveys). + + Args: + data: DataFrame containing the data. + predictors: List of predictor column names. + imputed_variables: List of variables to impute. + model_class: The Imputer class to use for evaluation. + ordering: How to order predictor removal: + - "mutual_info": Remove most informative first (default) + - "correlation": Remove highest correlated first + - "random": Remove in random order (control) + weight_col: Optional column name or array of sampling weights. + quantiles: List of quantiles for evaluation (default: [0.1, 0.3, 0.5, 0.7, 0.9]). + train_size: Proportion of data to use for training. + random_state: Random state for reproducibility. + + Returns: + Dictionary containing: + - results_df: DataFrame with columns ['step', 'predictor_removed', + 'remaining_predictors', 'quantile_loss', 'normalized_loss'] + - predictor_order: List of predictors in removal order + - baseline_loss: Performance with all predictors + - sensitivity_score: Area under degradation curve (higher = more sensitive) + """ + if quantiles is None: + quantiles = CIA_QUANTILES + + # Split data + train_data, test_data = train_test_split( + data, train_size=train_size, random_state=random_state + ) + + # Order predictors by importance + predictor_order = _order_predictors( + train_data, predictors, imputed_variables, ordering, random_state + ) + + # Compute baseline performance with all predictors + log.info("Computing baseline performance with all predictors") + baseline_loss = _evaluate_model( + train_data=train_data, + test_data=test_data, + predictors=predictors, + imputed_variables=imputed_variables, + model_class=model_class, + weight_col=weight_col, + quantiles=quantiles, + ) + + # Track results + results = [] + current_predictors = predictors.copy() + + # Add baseline (step 0, no predictors removed) + results.append( + { + "step": 0, + "predictor_removed": None, + "remaining_predictors": current_predictors.copy(), + "num_predictors": len(current_predictors), + "quantile_loss": baseline_loss, + "normalized_loss": 1.0, + } + ) + + # Progressively remove predictors + for step, pred_to_remove in enumerate( + tqdm(predictor_order, desc="Progressive exclusion"), start=1 + ): + current_predictors = [ + p for p in current_predictors if p != pred_to_remove + ] + + if len(current_predictors) == 0: + # No predictors left - record as maximum degradation + results.append( + { + "step": step, + "predictor_removed": pred_to_remove, + "remaining_predictors": [], + "num_predictors": 0, + "quantile_loss": np.nan, + "normalized_loss": np.nan, + } + ) + break + + try: + loss = _evaluate_model( + train_data=train_data, + test_data=test_data, + predictors=current_predictors, + imputed_variables=imputed_variables, + model_class=model_class, + weight_col=weight_col, + quantiles=quantiles, + ) + + normalized_loss = ( + loss / baseline_loss if baseline_loss > 0 else np.nan + ) + + results.append( + { + "step": step, + "predictor_removed": pred_to_remove, + "remaining_predictors": current_predictors.copy(), + "num_predictors": len(current_predictors), + "quantile_loss": loss, + "normalized_loss": normalized_loss, + } + ) + + except Exception as e: + log.warning( + f"Failed to evaluate after removing {pred_to_remove}: {e}" + ) + results.append( + { + "step": step, + "predictor_removed": pred_to_remove, + "remaining_predictors": current_predictors.copy(), + "num_predictors": len(current_predictors), + "quantile_loss": np.nan, + "normalized_loss": np.nan, + } + ) + + results_df = pd.DataFrame(results) + + # Compute sensitivity score (AUC of normalized loss curve) + # Higher = more sensitive to predictor removal + valid_results = results_df[results_df["normalized_loss"].notna()] + if len(valid_results) > 1: + # Use trapezoidal rule for AUC calculation + x = valid_results["step"].values + y = valid_results["normalized_loss"].values + # Normalize x to [0, 1] range + x_norm = x / x.max() if x.max() > 0 else x + sensitivity_score = trapezoid(y, x_norm) + else: + sensitivity_score = 1.0 + + return { + "results_df": results_df, + "predictor_order": predictor_order, + "baseline_loss": baseline_loss, + "sensitivity_score": sensitivity_score, + } + + +def _order_predictors( + data: pd.DataFrame, + predictors: List[str], + imputed_variables: List[str], + ordering: str, + random_state: int, +) -> List[str]: + """Order predictors by importance for removal. + + Args: + data: Training data. + predictors: List of predictor names. + imputed_variables: Target variables. + ordering: Ordering method ("mutual_info", "correlation", "random"). + random_state: Random state. + + Returns: + List of predictors ordered by importance (most important first). + """ + if ordering == "random": + rng = np.random.RandomState(random_state) + order = predictors.copy() + rng.shuffle(order) + return order + + # Compute importance scores + importance_scores = {} + + for pred in predictors: + scores = [] + for target in imputed_variables: + # Get predictor values - encode categorical if needed + X_series = data[pred] + if X_series.dtype == "object" or str(X_series.dtype) == "category": + # Encode categorical predictor as numeric codes + X = pd.Categorical(X_series).codes.reshape(-1, 1) + else: + X = X_series.values.reshape(-1, 1) + + y = data[target].values + + # Handle missing values using pd.isna (works with all types) + mask = ~(pd.isna(X.flatten()) | pd.isna(y)) + X_clean = X[mask].astype(float) + y_clean = y[mask].astype(float) + + if len(X_clean) < 10: + continue + + if ordering == "mutual_info": + mi = mutual_info_regression( + X_clean, y_clean, random_state=random_state + )[0] + scores.append(mi) + elif ordering == "correlation": + corr = np.abs(np.corrcoef(X_clean.flatten(), y_clean)[0, 1]) + scores.append(corr if not np.isnan(corr) else 0) + + importance_scores[pred] = np.mean(scores) if scores else 0 + + # Sort by importance (highest first - these will be removed first) + sorted_preds = sorted( + importance_scores.keys(), + key=lambda x: importance_scores[x], + reverse=True, + ) + + return sorted_preds + + +def _evaluate_model( + train_data: pd.DataFrame, + test_data: pd.DataFrame, + predictors: List[str], + imputed_variables: List[str], + model_class: Type[Imputer], + weight_col: Optional[Union[str, np.ndarray, pd.Series]], + quantiles: List[float], +) -> float: + """Train a model and evaluate its quantile loss. + + Returns: + Mean quantile loss across all quantiles and variables. + """ + model_name = model_class.__name__ + + # Initialize and fit the model + model = model_class() + + # QuantReg needs quantiles at fit time + if model_name == "QuantReg": + fitted_model = model.fit( + X_train=train_data, + predictors=predictors, + imputed_variables=imputed_variables, + weight_col=weight_col, + quantiles=quantiles, + ) + else: + fitted_model = model.fit( + X_train=train_data, + predictors=predictors, + imputed_variables=imputed_variables, + weight_col=weight_col, + ) + + # Get predictions + predictions = fitted_model.predict(test_data, quantiles) + + # Compute quantile loss using the existing function from metrics + losses = [] + for q in quantiles: + if q not in predictions: + continue + for var in imputed_variables: + if var not in predictions[q].columns: + continue + + true_values = test_data[var].values + pred_values = predictions[q][var].values + + # Use existing quantile_loss function + loss_array = quantile_loss(q, true_values, pred_values) + losses.append(np.mean(loss_array)) + + return np.mean(losses) if losses else np.nan + + +def plot_cia_degradation_curves( + results: Dict[str, Dict[str, Any]], + title: Optional[str] = None, + save_path: Optional[str] = None, + figsize: tuple = (PLOT_CONFIG["width"], PLOT_CONFIG["height"]), + use_absolute_loss: bool = True, +) -> go.Figure: + """Plot CIA degradation curves for multiple methods. + + Args: + results: Dict mapping method name to progressive_predictor_exclusion results. + title: Plot title. + save_path: Path to save the figure. + figsize: Figure size as (width, height). + use_absolute_loss: If True (default), plot absolute quantile loss values. + If False, plot normalized loss (relative to each method's baseline). + + Returns: + Plotly figure object. + """ + fig = go.Figure() + + y_col = "quantile_loss" if use_absolute_loss else "normalized_loss" + + for method_name, method_results in results.items(): + df = method_results["results_df"] + + # Skip if results_df is empty or missing required column + if df.empty or y_col not in df.columns: + continue + + valid_df = df[df[y_col].notna()] + + if valid_df.empty: + continue + + color = METHOD_COLORS.get(method_name, "#999999") + baseline = method_results.get("baseline_loss", np.nan) + + # Build label with baseline info + label = f"{method_name}" + if not np.isnan(baseline): + label += f" (baseline={baseline:.4f})" + + fig.add_trace( + go.Scatter( + x=valid_df["step"], + y=valid_df[y_col], + mode="lines+markers", + name=label, + line=dict(color=color, width=2), + marker=dict(color=color, size=8), + ) + ) + + if title is None: + title = "CIA Sensitivity: Performance Degradation as Predictors are Removed" + + y_axis_title = ( + "Quantile Loss" + if use_absolute_loss + else "Normalized Quantile Loss (1.0 = baseline)" + ) + + fig.update_layout( + title=title, + xaxis_title="Number of Predictors Removed", + yaxis_title=y_axis_title, + height=figsize[1], + width=figsize[0], + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], + legend=dict( + yanchor="top", + y=0.99, + xanchor="left", + x=0.01, + ), + hovermode="x unified", + ) + + fig.update_xaxes( + showgrid=PLOT_CONFIG["showgrid_x"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + fig.update_yaxes( + showgrid=PLOT_CONFIG["showgrid_y"], + gridcolor=PLOT_CONFIG["gridcolor"], + showline=PLOT_CONFIG["showline"], + linecolor=PLOT_CONFIG["linecolor"], + ) + + if save_path: + fig.write_html(save_path) + + return fig + + +def create_benchmark_summary_table( + cv_results: Dict[str, Dict[str, Dict[str, Any]]], + cia_results: Optional[Dict[str, Dict[str, Dict[str, Any]]]] = None, +) -> pd.DataFrame: + """Create a summary table of benchmark results with rankings. + + Args: + cv_results: Dict mapping dataset name to cv_results from autoimpute. + cia_results: Optional dict mapping dataset name to CIA sensitivity results. + + Returns: + DataFrame with columns: Dataset, Best Method, and rank columns for each method. + """ + methods = ["QRF", "OLS", "QuantReg", "Matching", "MDN"] + rows = [] + + for dataset_name, dataset_cv_results in cv_results.items(): + row = {"Dataset": dataset_name} + + # Extract mean quantile loss for each method + losses = {} + for method in methods: + if method in dataset_cv_results: + method_data = dataset_cv_results[method] + if "quantile_loss" in method_data: + ql = method_data["quantile_loss"] + losses[method] = ql.get("mean_test", np.nan) + else: + losses[method] = np.nan + else: + losses[method] = np.nan + + # Compute ranks + valid_losses = {k: v for k, v in losses.items() if not np.isnan(v)} + if valid_losses: + sorted_methods = sorted( + valid_losses.keys(), key=lambda x: valid_losses[x] + ) + ranks = {m: i + 1 for i, m in enumerate(sorted_methods)} + + # Fill in ranks for methods with NaN losses + max_rank = len(valid_losses) + 1 + for method in methods: + if method not in ranks: + ranks[method] = max_rank + + row["Best Method"] = sorted_methods[0] + else: + ranks = {m: np.nan for m in methods} + row["Best Method"] = "N/A" + + # Add rank columns + for method in methods: + row[f"{method} Rank"] = ranks.get(method, np.nan) + row[f"{method} Loss"] = losses.get(method, np.nan) + + # Add CIA sensitivity if provided + if cia_results and dataset_name in cia_results: + for method in methods: + if method in cia_results[dataset_name]: + row[f"{method} CIA"] = cia_results[dataset_name][ + method + ].get("sensitivity_score", np.nan) + else: + row[f"{method} CIA"] = np.nan + + rows.append(row) + + df = pd.DataFrame(rows) + + # Add mean rank row + mean_row = {"Dataset": "Mean Rank", "Best Method": "-"} + for method in methods: + rank_col = f"{method} Rank" + if rank_col in df.columns: + mean_row[rank_col] = df[rank_col].mean() + loss_col = f"{method} Loss" + if loss_col in df.columns: + mean_row[loss_col] = df[loss_col].mean() + cia_col = f"{method} CIA" + if cia_col in df.columns: + mean_row[cia_col] = df[cia_col].mean() + + df = pd.concat([df, pd.DataFrame([mean_row])], ignore_index=True) + + return df + + +def create_benchmark_heatmap( + cv_results: Dict[str, Dict[str, Dict[str, Any]]], + wasserstein_results: Optional[Dict[str, Dict[str, float]]] = None, + title: Optional[str] = None, + save_path: Optional[str] = None, + figsize: Optional[tuple] = None, +) -> go.Figure: + """Create a heatmap showing quantile loss and Wasserstein distance. + + The heatmap uses: + - Method-specific colors (consistent per column) + - Row-type opacity: Q-Loss rows are darker (0.8), W-Dist rows lighter (0.5) + - Font weight/size to indicate performance: best = bold + larger + + Args: + cv_results: Dict mapping dataset name to cv_results from autoimpute. + wasserstein_results: Optional dict mapping dataset to {method: distance}. + title: Plot title. + save_path: Path to save the figure. + figsize: Figure size (width, height). Auto-calculated if None. + + Returns: + Plotly figure object. + """ + methods = ["QRF", "OLS", "QuantReg", "Matching", "MDN"] + datasets = list(cv_results.keys()) + + # Build data matrix with row type tracking + row_labels = [] + row_types = [] # 'qloss' or 'wdist' + data_matrix = [] + + for dataset in datasets: + dataset_cv = cv_results[dataset] + + # Quantile Loss row + ql_row = [] + for method in methods: + if method in dataset_cv: + val = dataset_cv[method] + # Handle both nested format and simple float format + if isinstance(val, dict) and "quantile_loss" in val: + # Nested format: cv_results[dataset][method]["quantile_loss"]["mean_test"] + ql_row.append( + val["quantile_loss"].get("mean_test", np.nan) + ) + elif isinstance(val, (int, float)): + # Simple format: cv_results[dataset][method] = float + ql_row.append(val) + else: + ql_row.append(np.nan) + else: + ql_row.append(np.nan) + row_labels.append(f"{dataset} (Q-Loss)") + row_types.append("qloss") + data_matrix.append(ql_row) + + # Wasserstein Distance row + if wasserstein_results and dataset in wasserstein_results: + wd_row = [] + for method in methods: + wd_row.append(wasserstein_results[dataset].get(method, np.nan)) + row_labels.append(f"{dataset} (W-Dist)") + row_types.append("wdist") + data_matrix.append(wd_row) + + data_array = np.array(data_matrix) + n_rows = len(row_labels) + n_cols = len(methods) + + # Compute ranks for each row (1 = best, higher = worse) + rank_matrix = np.zeros_like(data_array) + for i in range(len(data_array)): + row = data_array[i] + valid_mask = ~np.isnan(row) + if valid_mask.sum() > 0: + # Rank valid values (lower value = better = rank 1) + valid_vals = row[valid_mask] + ranks = np.argsort(np.argsort(valid_vals)) + 1 + rank_matrix[i, valid_mask] = ranks + # Set NaN positions to max rank + 1 + rank_matrix[i, ~valid_mask] = valid_mask.sum() + 1 + + # Create figure + fig = go.Figure() + + # Create cells using shapes (rectangles) + annotations = [] + + for i in range(n_rows): + row_type = row_types[i] + # Set alpha based on row type: Q-Loss darker, W-Dist lighter + alpha = 0.8 if row_type == "qloss" else 0.5 + + for j, method in enumerate(methods): + value = data_array[i, j] + rank = rank_matrix[i, j] + + base_color = METHOD_COLORS.get(method, "#999999") + r = int(base_color[1:3], 16) + g = int(base_color[3:5], 16) + b = int(base_color[5:7], 16) + + if np.isnan(value): + cell_color = "rgba(200, 200, 200, 0.3)" + text = "N/A" + font_size = 10 + font_weight = "normal" + else: + cell_color = f"rgba({r}, {g}, {b}, {alpha:.2f})" + text = f"{value:.4f}" + # Best performer (rank 1) gets bold + larger font + if rank == 1: + font_size = 12 + font_weight = "bold" + else: + font_size = 10 + font_weight = "normal" + + # Add rectangle shape for cell background + fig.add_shape( + type="rect", + x0=j - 0.5, + x1=j + 0.5, + y0=i - 0.5, + y1=i + 0.5, + fillcolor=cell_color, + line=dict(color="white", width=1), + ) + + # Text color: white for MDN (dark indigo), black for others + if method == "MDN" and not np.isnan(value): + text_color = "white" + else: + text_color = "black" + + # Add text annotation with bold formatting for best + if font_weight == "bold": + text = f"{text}" + + annotations.append( + dict( + x=j, + y=i, + text=text, + showarrow=False, + font=dict(color=text_color, size=font_size), + xanchor="center", + yanchor="middle", + ) + ) + + # Add method color legend + for method in methods: + fig.add_trace( + go.Scatter( + x=[None], + y=[None], + mode="markers", + marker=dict( + size=12, color=METHOD_COLORS.get(method, "#999999") + ), + name=method, + ) + ) + + if title is None: + title = "Cross-Dataset Benchmark Results" + + # Calculate compact figure size if not provided + if figsize is None: + width = max(500, n_cols * 100 + 200) + height = max(300, n_rows * 40 + 100) + figsize = (width, height) + + fig.update_layout( + title=dict(text=title, font=dict(size=14)), + xaxis=dict( + tickmode="array", + tickvals=list(range(n_cols)), + ticktext=methods, + side="top", + showgrid=False, + zeroline=False, + range=[-0.5, n_cols - 0.5], + ), + yaxis=dict( + tickmode="array", + tickvals=list(range(n_rows)), + ticktext=row_labels, + showgrid=False, + zeroline=False, + autorange="reversed", + range=[-0.5, n_rows - 0.5], + ), + height=figsize[1], + width=figsize[0], + paper_bgcolor=PLOT_CONFIG["paper_bgcolor"], + plot_bgcolor=PLOT_CONFIG["plot_bgcolor"], + annotations=annotations, + legend=dict( + orientation="h", + yanchor="top", + y=-0.02, + xanchor="center", + x=0.5, + ), + margin=dict(l=180, r=20, t=60, b=40), + ) + + if save_path: + fig.write_html(save_path) + + return fig + + +def run_cia_analysis_for_dataset( + data: pd.DataFrame, + predictors: List[str], + imputed_variables: List[str], + model_classes: Optional[List[Type[Imputer]]] = None, + ordering: str = "mutual_info", + train_size: float = TRAIN_SIZE, + random_state: int = RANDOM_STATE, +) -> Dict[str, Dict[str, Any]]: + """Run CIA sensitivity analysis for all models on a dataset. + + Args: + data: DataFrame containing the data. + predictors: List of predictor column names. + imputed_variables: List of variables to impute. + model_classes: List of Imputer classes to evaluate. + ordering: Predictor ordering method. + train_size: Proportion for training. + random_state: Random state. + + Returns: + Dict mapping method name to progressive_predictor_exclusion results. + """ + if model_classes is None: + from microimpute.models import OLS, QRF, QuantReg + + model_classes = [QRF, OLS, QuantReg] + try: + from microimpute.models import Matching + + model_classes.append(Matching) + except ImportError: + pass + try: + from microimpute.models import MDN + + model_classes.append(MDN) + except ImportError: + pass + + results = {} + + for model_class in model_classes: + method_name = model_class.__name__ + + try: + method_results = progressive_predictor_exclusion( + data=data, + predictors=predictors, + imputed_variables=imputed_variables, + model_class=model_class, + ordering=ordering, + train_size=train_size, + random_state=random_state, + ) + results[method_name] = method_results + print( + f" {method_name} sensitivity score: " + f"{method_results['sensitivity_score']:.3f}" + ) + except Exception as e: + log.warning(f"CIA analysis failed for {method_name}: {e}") + print(f" CIA analysis failed for {method_name}: {e}") + results[method_name] = { + "results_df": pd.DataFrame(), + "predictor_order": [], + "baseline_loss": np.nan, + "sensitivity_score": np.nan, + } + + return results diff --git a/paper/benchmarking_datasets/data_loader.py b/paper/benchmarking_datasets/data_loader.py new file mode 100644 index 0000000..32421bd --- /dev/null +++ b/paper/benchmarking_datasets/data_loader.py @@ -0,0 +1,583 @@ +""" +OpenML AutoML Benchmark Regression Suite Data Loader + +This module retrieves and analyzes datasets from the OpenML AutoML +Benchmark Regression suite (ID: 269) for evaluating statistical +matching models in microimpute. +""" + +import re + +import openml +import pandas as pd + +# Patterns that indicate generic/non-interpretable feature names +GENERIC_FEATURE_PATTERNS = [ + r"^col_?\d+$", # col_1, col1, col_2 + r"^[fvx]\d+$", # f1, v1, x1, V1, X1 + r"^feat(ure)?_?\d+$", # feat1, feature_1 + r"^var_?\d+$", # var1, var_1 + r"^att(r)?_?\d+$", # attr1, att_1 + r"^[a-z]\d+[a-z]*\d*$", # P1, P1p2, H2p2 (encoded names) + r"^[a-f0-9]{8,}$", # Hashed/anonymized names (e.g., 48df886f9) +] + + +def is_generic_feature_name(name: str) -> bool: + """ + Check if a feature name appears to be generic/non-interpretable. + + Parameters + ---------- + name : str + Feature name to check + + Returns + ------- + bool + True if the name matches a generic pattern + """ + name_lower = name.lower().strip() + for pattern in GENERIC_FEATURE_PATTERNS: + if re.match(pattern, name_lower): + return True + return False + + +def calculate_interpretability_score(feature_names: list) -> float: + """ + Calculate the proportion of features with interpretable names. + + Parameters + ---------- + feature_names : list + List of feature names + + Returns + ------- + float + Proportion of features with non-generic names (0 to 1) + """ + if not feature_names: + return 0.0 + + interpretable_count = sum( + 1 for name in feature_names if not is_generic_feature_name(name) + ) + return interpretable_count / len(feature_names) + + +def list_all_qualities(dataset_id: int = 269) -> list: + """ + List all available qualities for a sample dataset to understand + what metadata OpenML provides. + + Parameters + ---------- + dataset_id : int + Sample dataset ID to inspect qualities + + Returns + ------- + list + List of all available quality names + """ + dataset = openml.datasets.get_dataset( + dataset_id, + download_data=False, + download_qualities=True, + ) + if dataset.qualities: + return sorted(dataset.qualities.keys()) + return [] + + +def get_benchmark_suite_metadata(suite_id: int = 269) -> pd.DataFrame: + """ + Retrieve metadata for all datasets in an OpenML benchmark suite. + + Parameters + ---------- + suite_id : int + The OpenML benchmark suite ID (default: 269 for AutoML + Benchmark Regression) + + Returns + ------- + pd.DataFrame + DataFrame containing metadata for each dataset in the suite + """ + # Get the benchmark suite + suite = openml.study.get_suite(suite_id) + print(f"Suite: {suite.name}") + print(f"Description: {suite.description}") + print(f"Number of datasets: {len(suite.data)}") + print("-" * 80) + + # Collect metadata for each dataset + metadata_list = [] + + for dataset_id in suite.data: + try: + dataset = openml.datasets.get_dataset( + dataset_id, + download_data=False, + download_qualities=True, + download_features_meta_data=True, + ) + + # Get qualities (dataset characteristics) + qualities = dataset.qualities if dataset.qualities else {} + + # Get feature information + features = dataset.features if dataset.features else {} + feature_names = [] + feature_types = {} + if features: + for feat in features.values(): + feature_names.append(feat.name) + feat_type = feat.data_type + feature_types[feat_type] = ( + feature_types.get(feat_type, 0) + 1 + ) + + # Calculate interpretability score + interpretability = calculate_interpretability_score(feature_names) + + metadata = { + "dataset_id": dataset_id, + "name": dataset.name, + "version": dataset.version, + "n_instances": qualities.get("NumberOfInstances"), + "n_features": qualities.get("NumberOfFeatures"), + "n_numeric_features": qualities.get("NumberOfNumericFeatures"), + "n_categorical_features": qualities.get( + "NumberOfSymbolicFeatures" + ), + "n_missing_values": qualities.get("NumberOfMissingValues"), + "pct_missing": qualities.get("PercentageOfMissingValues"), + "n_instances_with_missing": qualities.get( + "NumberOfInstancesWithMissingValues" + ), + # Relationship/correlation measures + "mean_mutual_info": qualities.get("MeanMutualInformation"), + "mean_attr_entropy": qualities.get("MeanAttributeEntropy"), + "equiv_num_attr": qualities.get("EquivalentNumberOfAtts"), + "noise_signal_ratio": qualities.get("NoiseToSignalRatio"), + "class_entropy": qualities.get("ClassEntropy"), + "mean_kurtosis": qualities.get("MeanKurtosisOfNumericAtts"), + "mean_skewness": qualities.get("MeanSkewnessOfNumericAtts"), + "target_variable": dataset.default_target_attribute, + "interpretability_score": interpretability, + "feature_names": feature_names, + "description": ( + dataset.description[:200] + "..." + if dataset.description and len(dataset.description) > 200 + else dataset.description + ), + "format": dataset.format, + "upload_date": dataset.upload_date, + } + + metadata_list.append(metadata) + print(f"Loaded: {dataset.name} (ID: {dataset_id})") + + except Exception as e: + print(f"Error loading dataset {dataset_id}: {e}") + metadata_list.append( + { + "dataset_id": dataset_id, + "name": "ERROR", + "error": str(e), + } + ) + + return pd.DataFrame(metadata_list) + + +def analyze_missingness(df_metadata: pd.DataFrame) -> pd.DataFrame: + """ + Analyze missingness patterns in the benchmark datasets. + + Parameters + ---------- + df_metadata : pd.DataFrame + DataFrame with dataset metadata + + Returns + ------- + pd.DataFrame + Summary of datasets with missing values + """ + missing_df = df_metadata[df_metadata["n_missing_values"] > 0].copy() + missing_df = missing_df.sort_values("pct_missing", ascending=False) + return missing_df[ + [ + "name", + "n_instances", + "n_features", + "n_missing_values", + "pct_missing", + "n_instances_with_missing", + ] + ] + + +def filter_datasets( + df_metadata: pd.DataFrame, + min_numeric_proportion: float = 0.5, + max_missing_values: int = 0, + max_instances: int = 40000, + min_instances: int = 0, + min_features: int = 0, + max_features: int = None, + exclude_name_patterns: list = None, + min_interpretability_score: float = None, +) -> pd.DataFrame: + """ + Filter datasets based on specified criteria. + + Parameters + ---------- + df_metadata : pd.DataFrame + DataFrame with dataset metadata from get_benchmark_suite_metadata() + min_numeric_proportion : float + Minimum proportion of features that must be numeric (default: 0.5) + max_missing_values : int + Maximum number of missing values allowed (default: 0 for complete) + max_instances : int + Maximum number of instances (default: 40000) + min_instances : int + Minimum number of instances (default: 0) + min_features : int + Minimum number of features (default: 0) + max_features : int + Maximum number of features (default: None for no limit) + exclude_name_patterns : list + List of substrings to exclude from dataset names (default: None) + Case-insensitive matching. + min_interpretability_score : float + Minimum proportion of features with interpretable names (default: + None for no filtering). Value between 0 and 1, where 1 means all + features must have interpretable names. + + Returns + ------- + pd.DataFrame + Filtered DataFrame containing only datasets meeting all criteria + """ + df = df_metadata.copy() + + # Calculate numeric proportion + df["numeric_proportion"] = df["n_numeric_features"] / df["n_features"] + + # Apply filters + mask = ( + (df["numeric_proportion"] >= min_numeric_proportion) + & (df["n_missing_values"] <= max_missing_values) + & (df["n_instances"] <= max_instances) + & (df["n_instances"] >= min_instances) + & (df["n_features"] >= min_features) + ) + + if max_features is not None: + mask = mask & (df["n_features"] <= max_features) + + # Filter by interpretability score + if min_interpretability_score is not None: + mask = mask & ( + df["interpretability_score"] >= min_interpretability_score + ) + + # Exclude datasets by name pattern + if exclude_name_patterns: + for pattern in exclude_name_patterns: + mask = mask & ( + ~df["name"].str.lower().str.contains(pattern.lower()) + ) + + filtered_df = df[mask].copy() + + return filtered_df + + +def save_filtered_dataset_ids( + df_filtered: pd.DataFrame, + output_path: str = "paper/benchmarking_datasets/selected_dataset_ids.txt", +) -> list: + """ + Save filtered dataset IDs to a file. + + Parameters + ---------- + df_filtered : pd.DataFrame + Filtered DataFrame from filter_datasets() + output_path : str + Path to save the dataset IDs + + Returns + ------- + list + List of selected dataset IDs + """ + dataset_ids = df_filtered["dataset_id"].tolist() + + with open(output_path, "w") as f: + f.write("# Selected OpenML Dataset IDs\n") + f.write("# Filtering criteria applied - see data_loader.py\n") + for did in dataset_ids: + f.write(f"{int(did)}\n") + + return dataset_ids + + +def download_filtered_datasets( + df_filtered: pd.DataFrame, + output_dir: str = "paper/benchmarking_datasets/datasets", + file_format: str = "csv", +) -> dict: + """ + Download all datasets that passed filtering criteria. + + Parameters + ---------- + df_filtered : pd.DataFrame + Filtered DataFrame from filter_datasets() + output_dir : str + Directory to save downloaded datasets + file_format : str + Output format: 'csv' or 'parquet' (default: 'csv') + + Returns + ------- + dict + Dictionary with dataset names as keys and info dict as values + containing 'path', 'shape', 'target', 'features' + """ + import os + + os.makedirs(output_dir, exist_ok=True) + + downloaded = {} + total = len(df_filtered) + + for _, row in df_filtered.iterrows(): + dataset_id = int(row["dataset_id"]) + dataset_name = row["name"] + + print( + f"[{len(downloaded) + 1}/{total}] Downloading {dataset_name} " + f"(ID: {dataset_id})..." + ) + + try: + dataset = openml.datasets.get_dataset( + dataset_id, download_data=True + ) + X, y, categorical_indicator, attribute_names = dataset.get_data( + target=dataset.default_target_attribute + ) + + # Combine features and target into single DataFrame + df = X.copy() + target_name = dataset.default_target_attribute + if target_name: + df[target_name] = y + + # Clean filename + clean_name = ( + dataset_name.lower().replace(" ", "_").replace("-", "_") + ) + + # Save to file + if file_format == "parquet": + output_path = os.path.join(output_dir, f"{clean_name}.parquet") + df.to_parquet(output_path, index=False) + else: + output_path = os.path.join(output_dir, f"{clean_name}.csv") + df.to_csv(output_path, index=False) + + downloaded[dataset_name] = { + "dataset_id": dataset_id, + "path": output_path, + "shape": df.shape, + "target": target_name, + "features": list(X.columns), + "categorical_features": ( + [ + name + for name, is_cat in zip( + attribute_names, categorical_indicator + ) + if is_cat + ] + if categorical_indicator + else [] + ), + } + + print( + f" Saved: {output_path} ({df.shape[0]} rows, " + f"{df.shape[1]} cols)" + ) + + except Exception as e: + print(f" ERROR: {e}") + downloaded[dataset_name] = {"error": str(e)} + + # Save summary + summary_path = os.path.join(output_dir, "download_summary.csv") + summary_data = [] + for name, info in downloaded.items(): + if "error" not in info: + summary_data.append( + { + "name": name, + "dataset_id": info["dataset_id"], + "path": info["path"], + "n_rows": info["shape"][0], + "n_cols": info["shape"][1], + "target": info["target"], + "n_categorical": len(info["categorical_features"]), + } + ) + pd.DataFrame(summary_data).to_csv(summary_path, index=False) + print(f"\nSummary saved to: {summary_path}") + print(f"Successfully downloaded: {len(summary_data)}/{total} datasets") + + return downloaded + + +def get_dataset_details(dataset_id: int) -> dict: + """ + Get detailed information about a specific dataset including + feature relationships. + + Parameters + ---------- + dataset_id : int + OpenML dataset ID + + Returns + ------- + dict + Detailed dataset information + """ + dataset = openml.datasets.get_dataset( + dataset_id, + download_data=True, + download_qualities=True, + download_features_meta_data=True, + ) + + X, _, categorical_indicator, attribute_names = dataset.get_data( + target=dataset.default_target_attribute + ) + + details = { + "name": dataset.name, + "description": dataset.description, + "target": dataset.default_target_attribute, + "features": attribute_names, + "categorical_features": [ + name + for name, is_cat in zip(attribute_names, categorical_indicator) + if is_cat + ], + "numeric_features": [ + name + for name, is_cat in zip(attribute_names, categorical_indicator) + if not is_cat + ], + "shape": X.shape, + "missing_per_feature": X.isnull().sum().to_dict(), + "data_sample": X.head(), + } + + return details + + +if __name__ == "__main__": + # Use first dataset in suite to discover available qualities + suite = openml.study.get_suite(269) + first_dataset_id = suite.data[0] + all_qualities = list_all_qualities(first_dataset_id) + + # Fetch and display benchmark suite metadata + print("\n" + "=" * 80) + print("OpenML AutoML Benchmark Regression Suite Analysis") + print("=" * 80) + + metadata_df = get_benchmark_suite_metadata(269) + + # Display summary table + print("\n" + "=" * 80) + print("DATASET SUMMARY TABLE - Basic Info") + print("=" * 80) + + # Set display options for better viewing + pd.set_option("display.max_columns", None) + pd.set_option("display.width", None) + pd.set_option("display.max_colwidth", 50) + + summary_cols = [ + "dataset_id", + "name", + "n_instances", + "n_features", + "n_numeric_features", + "n_categorical_features", + ] + + print(metadata_df[summary_cols].to_string(index=False)) + + # Filter datasets based on criteria + print("\n" + "=" * 80) + print("FILTERED DATASETS FOR BENCHMARKING") + print("=" * 80) + + # Define filtering parameters + FILTER_PARAMS = { + "min_numeric_proportion": 0.5, # At least 50% numeric features + "max_missing_values": 0, # No missing values + "max_instances": 40000, # Less than 40k instances + "min_instances": 1000, # At least 1k instances + "min_features": 0, + "max_features": None, + "exclude_name_patterns": ["QSAR", "MIP-2016", "yprop", "topo", "wine"], + "min_interpretability_score": 0.8, # 80% interpretable features + } + + print("\nFiltering criteria:") + for param, value in FILTER_PARAMS.items(): + print(f" {param}: {value}") + + filtered_df = filter_datasets(metadata_df, **FILTER_PARAMS) + + print( + f"\nDatasets meeting criteria: {len(filtered_df)} / {len(metadata_df)}" + ) + print("\n" + "-" * 80) + + filter_cols = [ + "dataset_id", + "name", + "n_instances", + "n_features", + "n_numeric_features", + "numeric_proportion", + "interpretability_score", + ] + + print(filtered_df[filter_cols].to_string(index=False)) + + # Print filtered dataset IDs + selected_ids = filtered_df["dataset_id"].tolist() + print(f"IDs: {selected_ids}") + + # Download all filtered datasets + downloaded = download_filtered_datasets( + filtered_df, + output_dir="paper/benchmarking_datasets/datasets", + file_format="csv", + ) diff --git a/paper/benchmarking_datasets/dataset_analysis.ipynb b/paper/benchmarking_datasets/dataset_analysis.ipynb new file mode 100644 index 0000000..19e57fb --- /dev/null +++ b/paper/benchmarking_datasets/dataset_analysis.ipynb @@ -0,0 +1,29933 @@ +{ + "cells": [ + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "# Benchmarking Datasets Analysis\n", + "\n", + "This notebook analyzes each dataset in the benchmarking datasets folder:\n", + "1. Computes predictor correlations (Pearson, Spearman, and Mutual Information)\n", + "2. Runs leave-one-out analysis to identify important predictors\n", + "3. Creates 60/40 train/test splits for donor/receiver data\n", + "4. Runs autoimpute using the donor/receiver splits" + ] + }, + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [], + "source": [ + "import os\n", + "import pandas as pd\n", + "import numpy as np\n", + "from sklearn.model_selection import train_test_split\n", + "from microimpute import (\n", + " method_comparison_results,\n", + " compute_predictor_correlations,\n", + " leave_one_out_analysis,\n", + " autoimpute,\n", + " QRF,\n", + " QuantReg, \n", + " OLS, \n", + " Matching,\n", + ")\n", + "from microimpute.comparisons import compare_distributions\n", + "\n", + "# Import benchmark utilities for CIA analysis and cross-dataset summary\n", + "from benchmark_utils import (\n", + " progressive_predictor_exclusion,\n", + " plot_cia_degradation_curves,\n", + " create_benchmark_summary_table,\n", + " create_benchmark_heatmap,\n", + " run_cia_analysis_for_dataset,\n", + ")\n", + "\n", + "pd.set_option('display.max_columns', None)\n", + "pd.set_option('display.width', None)\n", + "pd.set_option('display.max_colwidth', 50)" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Available datasets:\n", + " name target n_rows n_cols\n", + "0 space_ga ln(VOTES/POP) 3107 7\n", + "1 wine_quality quality 6497 12\n", + "2 elevators Goal 16599 19\n", + "3 Brazilian_houses total_(BRL) 10692 13\n", + "4 OnlineNewsPopularity shares 39644 60\n", + "5 abalone Class_number_of_rings 4177 9\n", + "6 house_sales price 21613 22\n" + ] + } + ], + "source": [ + "DATASETS_DIR = \"datasets\"\n", + "\n", + "summary_df = pd.read_csv(os.path.join(DATASETS_DIR, \"download_summary.csv\"))\n", + "print(\"Available datasets:\")\n", + "print(summary_df[[\"name\", \"target\", \"n_rows\", \"n_cols\"]])" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [], + "source": [ + "def load_dataset(name):\n", + " \"\"\"Load a dataset by name and return the DataFrame.\"\"\"\n", + " # Clean name to match file naming convention\n", + " clean_name = name.lower().replace(\" \", \"_\").replace(\"-\", \"_\")\n", + " file_path = os.path.join(DATASETS_DIR, f\"{clean_name}.csv\")\n", + " return pd.read_csv(file_path)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "---\n", + "## 1. Space GA Dataset\n", + "\n", + "Georgia county voting data with geographic variables." + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "Shape: (3107, 7)\n", + "\n", + "Column names:\n", + " 1. POP\n", + " 2. EDUCATION\n", + " 3. HOUSES\n", + " 4. INCOME\n", + " 5. XCOORD\n", + " 6. YCOORD\n", + " 7. ln(VOTES/POP)\n", + "\n", + "Data types:\n", + "POP float64\n", + "EDUCATION float64\n", + "HOUSES float64\n", + "INCOME float64\n", + "XCOORD float64\n", + "YCOORD float64\n", + "ln(VOTES/POP) float64\n", + "dtype: object\n", + "\n", + "First few rows:\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
POPEDUCATIONHOUSESINCOMEXCOORDYCOORDln(VOTES/POP)
09.9729209.2462869.00405412.134915-86641472.032542207.0-0.661559
110.90334710.2212149.96575813.056638-87754736.030654881.0-0.650859
29.7222058.7535298.70764811.630628-85388993.031863073.0-0.617114
39.2736918.1831188.27741211.243712-87126855.032996943.0-0.639070
410.1515199.2077379.24067612.155100-86566214.033979740.0-0.700274
\n", + "
" + ], + "text/plain": [ + " POP EDUCATION HOUSES INCOME XCOORD YCOORD \\\n", + "0 9.972920 9.246286 9.004054 12.134915 -86641472.0 32542207.0 \n", + "1 10.903347 10.221214 9.965758 13.056638 -87754736.0 30654881.0 \n", + "2 9.722205 8.753529 8.707648 11.630628 -85388993.0 31863073.0 \n", + "3 9.273691 8.183118 8.277412 11.243712 -87126855.0 32996943.0 \n", + "4 10.151519 9.207737 9.240676 12.155100 -86566214.0 33979740.0 \n", + "\n", + " ln(VOTES/POP) \n", + "0 -0.661559 \n", + "1 -0.650859 \n", + "2 -0.617114 \n", + "3 -0.639070 \n", + "4 -0.700274 " + ] + }, + "execution_count": 4, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_space_ga = load_dataset(\"space_ga\")\n", + "print(f\"\\nShape: {df_space_ga.shape}\")\n", + "print(f\"\\nColumn names:\")\n", + "for i, col in enumerate(df_space_ga.columns):\n", + " print(f\" {i+1}. {col}\")\n", + "print(f\"\\nData types:\")\n", + "print(df_space_ga.dtypes)\n", + "print(f\"\\nFirst few rows:\")\n", + "df_space_ga.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 5, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: ln(VOTES/POP)\n", + "Predictors (6): ['POP', 'EDUCATION', 'HOUSES', 'INCOME', 'XCOORD', 'YCOORD']\n" + ] + } + ], + "source": [ + "space_ga_target = \"ln(VOTES/POP)\" # Target variable to impute\n", + "space_ga_predictors = [col for col in df_space_ga.columns if col != space_ga_target]\n", + "\n", + "print(f\"Target variable: {space_ga_target}\")\n", + "print(f\"Predictors ({len(space_ga_predictors)}): {space_ga_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 6, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
POPEDUCATIONHOUSESINCOMEXCOORDYCOORD
POP1.0000000.9874720.9940020.9928990.267719-0.040601
EDUCATION0.9874721.0000000.9843650.9931330.1991290.047424
HOUSES0.9940020.9843651.0000000.9880840.268707-0.030783
INCOME0.9928990.9931330.9880841.0000000.231344-0.008157
XCOORD0.2677190.1991290.2687070.2313441.000000-0.193174
YCOORD-0.0406010.047424-0.030783-0.008157-0.1931741.000000
\n", + "
" + ], + "text/plain": [ + " POP EDUCATION HOUSES INCOME XCOORD YCOORD\n", + "POP 1.000000 0.987472 0.994002 0.992899 0.267719 -0.040601\n", + "EDUCATION 0.987472 1.000000 0.984365 0.993133 0.199129 0.047424\n", + "HOUSES 0.994002 0.984365 1.000000 0.988084 0.268707 -0.030783\n", + "INCOME 0.992899 0.993133 0.988084 1.000000 0.231344 -0.008157\n", + "XCOORD 0.267719 0.199129 0.268707 0.231344 1.000000 -0.193174\n", + "YCOORD -0.040601 0.047424 -0.030783 -0.008157 -0.193174 1.000000" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
POPEDUCATIONHOUSESINCOMEXCOORDYCOORD
POP1.0000000.9822080.9942490.9907240.353281-0.022140
EDUCATION0.9822081.0000000.9809170.9910550.2809710.082580
HOUSES0.9942490.9809171.0000000.9870130.350405-0.010133
INCOME0.9907240.9910550.9870131.0000000.3163620.020701
XCOORD0.3532810.2809710.3504050.3163621.000000-0.139110
YCOORD-0.0221400.082580-0.0101330.020701-0.1391101.000000
\n", + "
" + ], + "text/plain": [ + " POP EDUCATION HOUSES INCOME XCOORD YCOORD\n", + "POP 1.000000 0.982208 0.994249 0.990724 0.353281 -0.022140\n", + "EDUCATION 0.982208 1.000000 0.980917 0.991055 0.280971 0.082580\n", + "HOUSES 0.994249 0.980917 1.000000 0.987013 0.350405 -0.010133\n", + "INCOME 0.990724 0.991055 0.987013 1.000000 0.316362 0.020701\n", + "XCOORD 0.353281 0.280971 0.350405 0.316362 1.000000 -0.139110\n", + "YCOORD -0.022140 0.082580 -0.010133 0.020701 -0.139110 1.000000" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
POPEDUCATIONHOUSESINCOMEXCOORDYCOORD
POP1.0000000.1733670.2299310.1879040.0180960.004794
EDUCATION0.1733671.0000000.1672370.1955080.0163090.004089
HOUSES0.2299310.1672371.0000000.1782650.0198290.006102
INCOME0.1879040.1955080.1782651.0000000.0152910.006218
XCOORD0.0180960.0163090.0198290.0152911.0000000.000000
YCOORD0.0047940.0040890.0061020.0062180.0000001.000000
\n", + "
" + ], + "text/plain": [ + " POP EDUCATION HOUSES INCOME XCOORD YCOORD\n", + "POP 1.000000 0.173367 0.229931 0.187904 0.018096 0.004794\n", + "EDUCATION 0.173367 1.000000 0.167237 0.195508 0.016309 0.004089\n", + "HOUSES 0.229931 0.167237 1.000000 0.178265 0.019829 0.006102\n", + "INCOME 0.187904 0.195508 0.178265 1.000000 0.015291 0.006218\n", + "XCOORD 0.018096 0.016309 0.019829 0.015291 1.000000 0.000000\n", + "YCOORD 0.004794 0.004089 0.006102 0.006218 0.000000 1.000000" + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
ln(VOTES/POP)
YCOORD0.021069
XCOORD0.017490
POP0.012771
HOUSES0.011503
INCOME0.010931
EDUCATION0.009178
\n", + "
" + ], + "text/plain": [ + " ln(VOTES/POP)\n", + "YCOORD 0.021069\n", + "XCOORD 0.017490\n", + "POP 0.012771\n", + "HOUSES 0.011503\n", + "INCOME 0.010931\n", + "EDUCATION 0.009178" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "space_ga_correlations = compute_predictor_correlations(\n", + " data=df_space_ga,\n", + " predictors=space_ga_predictors,\n", + " imputed_variables=[space_ga_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(space_ga_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(space_ga_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(space_ga_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(space_ga_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=space_ga_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 7, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "261f3646fa2b4cdfa55c0aa2a4da3e7e", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/6 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
5YCOORD0.04789200.00861321.9271960.0392790
4XCOORD0.04501700.00573814.6084860.0392790
2HOUSES0.03993400.0006551.6669950.0392790
0POP0.03971000.0004311.0960660.0392790
1EDUCATION0.03936100.0000820.2084180.0392790
3INCOME0.0390250-0.000254-0.6471400.0392790
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "5 YCOORD 0.047892 0 0.008613 \n", + "4 XCOORD 0.045017 0 0.005738 \n", + "2 HOUSES 0.039934 0 0.000655 \n", + "0 POP 0.039710 0 0.000431 \n", + "1 EDUCATION 0.039361 0 0.000082 \n", + "3 INCOME 0.039025 0 -0.000254 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "5 21.927196 0.039279 0 \n", + "4 14.608486 0.039279 0 \n", + "2 1.666995 0.039279 0 \n", + "0 1.096066 0.039279 0 \n", + "1 0.208418 0.039279 0 \n", + "3 -0.647140 0.039279 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis...\")\n", + "space_ga_loo = leave_one_out_analysis(\n", + " data=df_space_ga,\n", + " predictors=space_ga_predictors,\n", + " imputed_variables=[space_ga_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(space_ga_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 1864\n", + "Receiver size: 1243\n", + "\n", + "Running autoimpute...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "9c2c432f784c44549683f59d501cfe0f", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.01817500633531141, + 0.026700310464725936, + 0.03332652926829131, + 0.03909195798332123, + 0.04481586833567268, + 0.04930413917960528, + 0.05246454656928429, + 0.052874904051341154, + 0.052681695396049846, + 0.05153127849087411, + 0.05240536885595963, + 0.04965082859710386, + 0.04572216335007338, + 0.04178626351653487, + 0.037213526486628165, + 0.032431151042838766, + 0.02616058441739506, + 0.020133327326486324, + 0.013122355392717968 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.0010833911164227247, + 0.0011706862196016997, + 0.0009853048017531005, + 0.000892210416357871, + 0.0009172973122542029, + 0.001164622135044908, + 0.0014272021984131826, + 0.001686676762208881, + 0.001831444691663165, + 0.0019462562032744573, + 0.00194753722126197, + 0.0019407490556580332, + 0.0018799036550497994, + 0.0017377065265933025, + 0.001591968134523271, + 0.0013560685519335932, + 0.0010980755543928862, + 0.0008820058771501513, + 0.0007627430667108099 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.013601495172351546, + 0.022039520319063255, + 0.028582415903448954, + 0.033730970460644716, + 0.03799893409026435, + 0.04147346522482332, + 0.04428975359762964, + 0.04629995838053462, + 0.047597906635362466, + 0.04809691002792157, + 0.047873756746492246, + 0.046863588384280196, + 0.04505972475506838, + 0.042536489732329034, + 0.03925222528048754, + 0.03509874687230753, + 0.02978220718670343, + 0.022938535366078407, + 0.01405148700683857 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.011188471044211747, + 0.011170466264537793, + 0.007572416709230087, + 0.008046557525503314, + 0.006599233742856614, + 0.005785321345071802, + 0.004151473977884341, + 0.0034082976292562793, + 0.00280840105564643, + 0.002763961428761023, + 0.0030371975282827667, + 0.003647703367441269, + 0.0039005773788334955, + 0.004922978653361434, + 0.005063572136316329, + 0.005799875185466762, + 0.006854700224737245, + 0.007405866165028072, + 0.008082787419597216 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.08281870777807214, + 0.08151629060086235, + 0.08802879746544752, + 0.08784592140541951, + 0.09231444655153817, + 0.09395360366830244, + 0.09430540688928825, + 0.09437480622748887, + 0.09371971036321496, + 0.09283779711962045, + 0.09115149545375179, + 0.08931422613245024, + 0.0865236553694542, + 0.0837545184989678, + 0.0780786632773851, + 0.07512390387644333, + 0.06999364643389736, + 0.06438587458499198, + 0.05699896802855233 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.004990500230487147, + 0.004500933987506405, + 0.004039452394904967, + 0.0036168217018443067, + 0.0032482422133479484, + 0.002954015980260504, + 0.0027580424268144484, + 0.0026819471146907023, + 0.0027357516892624442, + 0.0029122652636774096, + 0.0031911902651137842, + 0.0035484583035040856, + 0.003962937299109404, + 0.004418556599507702, + 0.004903862507335305, + 0.005410873057100847, + 0.005934027441012225, + 0.006469410389110239, + 0.007014222307297095 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0662964790001646, + 0.06590329223614041, + 0.06551010547211625, + 0.06511691870809205, + 0.06472373194406786, + 0.06433054518004368, + 0.06393735841601951, + 0.06354417165199532, + 0.06315098488797113, + 0.06275779812394694, + 0.06236461135992276, + 0.06197142459589857, + 0.0615782378318744, + 0.0611850510678502, + 0.060791864303826014, + 0.060398677539801826, + 0.060005490775777645, + 0.05961230401175346, + 0.05921911724772928 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.0033196477456942774, + 0.004851665534085506, + 0.005580743734245581, + 0.00595074087756197, + 0.00507136013529618, + 0.004385024588997225, + 0.005090702063578518, + 0.00490024195318755, + 0.004839620212289013, + 0.004905549507001035, + 0.005263668304284575, + 0.005931056011551634, + 0.006897960128047957, + 0.008196001934724374, + 0.009316496895623303, + 0.00969934721819898, + 0.009198206736044214, + 0.00772767756159229, + 0.005294394589765286 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.022783922599610946, + 0.03559129951075892, + 0.04464404245361313, + 0.05051369261676569, + 0.05519761013291058, + 0.05817472426610678, + 0.059641272680632317, + 0.061052271278365086, + 0.061756079716894605, + 0.061930196518500166, + 0.06169980566566542, + 0.06096723283012295, + 0.0597625597927293, + 0.05776486409427974, + 0.05472986967183059, + 0.05020685073900701, + 0.04374091826469292, + 0.03486241920427831, + 0.02278261198030372 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.03892588447685344, + "y1": 0.03892588447685344 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.03616674163908578, + "y1": 0.03616674163908578 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.08405475998553413, + "y1": 0.08405475998553413 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.06275779812394694, + "y1": 0.06275779812394694 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.05041064442195096, + "y1": 0.05041064442195096 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Space GA Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "space_ga_donor, space_ga_receiver = train_test_split(\n", + " df_space_ga, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "# Remove target from receiver (simulating missing data)\n", + "space_ga_receiver_no_target = space_ga_receiver.drop(columns=[space_ga_target])\n", + "\n", + "print(f\"Donor size: {len(space_ga_donor)}\")\n", + "print(f\"Receiver size: {len(space_ga_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute...\")\n", + "space_ga_result = autoimpute(\n", + " donor_data=space_ga_donor,\n", + " receiver_data=space_ga_receiver_no_target.copy(),\n", + " predictors=space_ga_predictors,\n", + " imputed_variables=[space_ga_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Benchmarking Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=space_ga_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + " data_format=\"wide\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Space GA Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 9, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (Space GA) ---\n", + "OLS: Wasserstein Distance = 0.0272\n", + "QRF: Wasserstein Distance = 0.0101\n", + "QuantReg: Wasserstein Distance = 0.0981\n", + "Matching: Wasserstein Distance = 0.0111\n", + "MDN: Wasserstein Distance = 0.0504\n" + ] + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (Space GA) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "space_ga_wasserstein = {}\n", + "\n", + "distribution_comparison_space_ga = compare_distributions(\n", + " donor_data=space_ga_receiver, # Ground truth test split\n", + " receiver_data=space_ga_result.receiver_data, # Imputed values\n", + " imputed_variables=[space_ga_target],\n", + ")\n", + "best_method_name = space_ga_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "space_ga_wasserstein[best_method_name] = distribution_comparison_space_ga[\n", + " distribution_comparison_space_ga['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {space_ga_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in space_ga_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = space_ga_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [space_ga_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=space_ga_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[space_ga_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " space_ga_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {space_ga_wasserstein.get(method_name, 'N/A'):.4f}\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 10, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "3b2c921f8eab4a1e85a1930e593a3c22", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/6 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
climbRateSgzpqcurRollabsRolldiffClbdiffRollRatediffDiffClbSaTime1SaTime2SaTime3SaTime4diffSaTime1diffSaTime2diffSaTime3diffSaTime4SaGoal
0118.0-55.0-0.28-0.08-0.2-11.011.00.005-0.2-0.0010-0.0010-0.0010-0.00100.00000.00.00.0-0.00100.031
1390.0-45.0-0.06-0.07-0.6-12.011.00.010-0.2-0.0008-0.0008-0.0008-0.00080.00000.00.00.0-0.00080.034
268.06.00.110.150.6-10.0-9.0-0.003-0.2-0.0011-0.0010-0.0010-0.0010-0.00020.00.00.0-0.00100.033
3-358.0-12.0-0.200.13-0.3-11.0-7.00.001-0.1-0.0010-0.0010-0.0010-0.00100.00000.00.00.0-0.00100.032
4-411.0-19.0-0.180.02-0.5-11.0-3.00.0021.2-0.0010-0.0010-0.0010-0.00100.00000.00.00.0-0.00100.030
\n", + "" + ], + "text/plain": [ + " climbRate Sgz p q curRoll absRoll diffClb diffRollRate \\\n", + "0 118.0 -55.0 -0.28 -0.08 -0.2 -11.0 11.0 0.005 \n", + "1 390.0 -45.0 -0.06 -0.07 -0.6 -12.0 11.0 0.010 \n", + "2 68.0 6.0 0.11 0.15 0.6 -10.0 -9.0 -0.003 \n", + "3 -358.0 -12.0 -0.20 0.13 -0.3 -11.0 -7.0 0.001 \n", + "4 -411.0 -19.0 -0.18 0.02 -0.5 -11.0 -3.0 0.002 \n", + "\n", + " diffDiffClb SaTime1 SaTime2 SaTime3 SaTime4 diffSaTime1 diffSaTime2 \\\n", + "0 -0.2 -0.0010 -0.0010 -0.0010 -0.0010 0.0000 0.0 \n", + "1 -0.2 -0.0008 -0.0008 -0.0008 -0.0008 0.0000 0.0 \n", + "2 -0.2 -0.0011 -0.0010 -0.0010 -0.0010 -0.0002 0.0 \n", + "3 -0.1 -0.0010 -0.0010 -0.0010 -0.0010 0.0000 0.0 \n", + "4 1.2 -0.0010 -0.0010 -0.0010 -0.0010 0.0000 0.0 \n", + "\n", + " diffSaTime3 diffSaTime4 Sa Goal \n", + "0 0.0 0.0 -0.0010 0.031 \n", + "1 0.0 0.0 -0.0008 0.034 \n", + "2 0.0 0.0 -0.0010 0.033 \n", + "3 0.0 0.0 -0.0010 0.032 \n", + "4 0.0 0.0 -0.0010 0.030 " + ] + }, + "execution_count": 11, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_elevators = load_dataset(\"elevators\")\n", + "print(f\"\\nShape: {df_elevators.shape}\")\n", + "print(f\"\\nColumn names:\")\n", + "for i, col in enumerate(df_elevators.columns):\n", + " print(f\" {i+1}. {col}\")\n", + "print(f\"\\nData types:\")\n", + "print(df_elevators.dtypes)\n", + "print(f\"\\nFirst few rows:\")\n", + "df_elevators.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 12, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: Goal\n", + "Predictors (8): ['climbRate', 'Sgz', 'curRoll', 'absRoll', 'SaTime1', 'SaTime2', 'SaTime3', 'SaTime4']\n" + ] + } + ], + "source": [ + "elevators_target = \"Goal\" # Target variable to impute\n", + "elevators_predictors = [\"climbRate\", \"Sgz\", \"curRoll\", \"absRoll\", \"SaTime1\", \"SaTime2\", \"SaTime3\", \"SaTime4\"]\n", + "\n", + "print(f\"Target variable: {elevators_target}\")\n", + "print(f\"Predictors ({len(elevators_predictors)}): {elevators_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 13, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
climbRateSgzcurRollabsRollSaTime1SaTime2SaTime3SaTime4
climbRate1.000000-0.013900-0.146812-0.0234820.0947900.1040380.1040720.107007
Sgz-0.0139001.0000000.0170430.0771190.0622840.0703090.0703080.076198
curRoll-0.1468120.0170431.0000000.194465-0.230882-0.227539-0.227519-0.224023
absRoll-0.0234820.0771190.1944651.0000000.6831320.6789350.6789540.678385
SaTime10.0947900.062284-0.2308820.6831321.0000000.9880010.9879950.976332
SaTime20.1040380.070309-0.2275390.6789350.9880011.0000000.9999910.994399
SaTime30.1040720.070308-0.2275190.6789540.9879950.9999911.0000000.994408
SaTime40.1070070.076198-0.2240230.6783850.9763320.9943990.9944081.000000
\n", + "
" + ], + "text/plain": [ + " climbRate Sgz curRoll absRoll SaTime1 SaTime2 \\\n", + "climbRate 1.000000 -0.013900 -0.146812 -0.023482 0.094790 0.104038 \n", + "Sgz -0.013900 1.000000 0.017043 0.077119 0.062284 0.070309 \n", + "curRoll -0.146812 0.017043 1.000000 0.194465 -0.230882 -0.227539 \n", + "absRoll -0.023482 0.077119 0.194465 1.000000 0.683132 0.678935 \n", + "SaTime1 0.094790 0.062284 -0.230882 0.683132 1.000000 0.988001 \n", + "SaTime2 0.104038 0.070309 -0.227539 0.678935 0.988001 1.000000 \n", + "SaTime3 0.104072 0.070308 -0.227519 0.678954 0.987995 0.999991 \n", + "SaTime4 0.107007 0.076198 -0.224023 0.678385 0.976332 0.994399 \n", + "\n", + " SaTime3 SaTime4 \n", + "climbRate 0.104072 0.107007 \n", + "Sgz 0.070308 0.076198 \n", + "curRoll -0.227519 -0.224023 \n", + "absRoll 0.678954 0.678385 \n", + "SaTime1 0.987995 0.976332 \n", + "SaTime2 0.999991 0.994399 \n", + "SaTime3 1.000000 0.994408 \n", + "SaTime4 0.994408 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
climbRateSgzcurRollabsRollSaTime1SaTime2SaTime3SaTime4
climbRate1.000000-0.017436-0.147546-0.0199830.1186370.1277930.1277980.131181
Sgz-0.0174361.0000000.0143490.0887000.0569700.0654030.0654050.071326
curRoll-0.1475460.0143491.0000000.196730-0.257374-0.252452-0.252451-0.247426
absRoll-0.0199830.0887000.1967301.0000000.7061200.7006930.7006950.699823
SaTime10.1186370.056970-0.2573740.7061201.0000000.9860840.9860840.974161
SaTime20.1277930.065403-0.2524520.7006930.9860841.0000001.0000000.993382
SaTime30.1277980.065405-0.2524510.7006950.9860841.0000001.0000000.993382
SaTime40.1311810.071326-0.2474260.6998230.9741610.9933820.9933821.000000
\n", + "
" + ], + "text/plain": [ + " climbRate Sgz curRoll absRoll SaTime1 SaTime2 \\\n", + "climbRate 1.000000 -0.017436 -0.147546 -0.019983 0.118637 0.127793 \n", + "Sgz -0.017436 1.000000 0.014349 0.088700 0.056970 0.065403 \n", + "curRoll -0.147546 0.014349 1.000000 0.196730 -0.257374 -0.252452 \n", + "absRoll -0.019983 0.088700 0.196730 1.000000 0.706120 0.700693 \n", + "SaTime1 0.118637 0.056970 -0.257374 0.706120 1.000000 0.986084 \n", + "SaTime2 0.127793 0.065403 -0.252452 0.700693 0.986084 1.000000 \n", + "SaTime3 0.127798 0.065405 -0.252451 0.700695 0.986084 1.000000 \n", + "SaTime4 0.131181 0.071326 -0.247426 0.699823 0.974161 0.993382 \n", + "\n", + " SaTime3 SaTime4 \n", + "climbRate 0.127798 0.131181 \n", + "Sgz 0.065405 0.071326 \n", + "curRoll -0.252451 -0.247426 \n", + "absRoll 0.700695 0.699823 \n", + "SaTime1 0.986084 0.974161 \n", + "SaTime2 1.000000 0.993382 \n", + "SaTime3 1.000000 0.993382 \n", + "SaTime4 0.993382 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
climbRateSgzcurRollabsRollSaTime1SaTime2SaTime3SaTime4
climbRate1.0000000.0017570.0012540.0027380.0018810.0048530.0047760.006018
Sgz0.0017571.0000000.0000000.0043380.0030360.0036550.0036560.004861
curRoll0.0012540.0000001.0000000.0532930.0137250.0132560.0132140.012820
absRoll0.0027380.0043380.0532931.0000000.0952710.0936170.0935530.094038
SaTime10.0018810.0030360.0137250.0952711.0000000.5057160.5056460.458947
SaTime20.0048530.0036550.0132560.0936170.5057161.0000000.6917060.584177
SaTime30.0047760.0036560.0132140.0935530.5056460.6917061.0000000.584307
SaTime40.0060180.0048610.0128200.0940380.4589470.5841770.5843071.000000
\n", + "
" + ], + "text/plain": [ + " climbRate Sgz curRoll absRoll SaTime1 SaTime2 \\\n", + "climbRate 1.000000 0.001757 0.001254 0.002738 0.001881 0.004853 \n", + "Sgz 0.001757 1.000000 0.000000 0.004338 0.003036 0.003655 \n", + "curRoll 0.001254 0.000000 1.000000 0.053293 0.013725 0.013256 \n", + "absRoll 0.002738 0.004338 0.053293 1.000000 0.095271 0.093617 \n", + "SaTime1 0.001881 0.003036 0.013725 0.095271 1.000000 0.505716 \n", + "SaTime2 0.004853 0.003655 0.013256 0.093617 0.505716 1.000000 \n", + "SaTime3 0.004776 0.003656 0.013214 0.093553 0.505646 0.691706 \n", + "SaTime4 0.006018 0.004861 0.012820 0.094038 0.458947 0.584177 \n", + "\n", + " SaTime3 SaTime4 \n", + "climbRate 0.004776 0.006018 \n", + "Sgz 0.003656 0.004861 \n", + "curRoll 0.013214 0.012820 \n", + "absRoll 0.093553 0.094038 \n", + "SaTime1 0.505646 0.458947 \n", + "SaTime2 0.691706 0.584177 \n", + "SaTime3 1.000000 0.584307 \n", + "SaTime4 0.584307 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
Goal
SaTime10.055697
SaTime30.048387
SaTime20.048359
SaTime40.047626
absRoll0.025873
climbRate0.013537
Sgz0.005171
curRoll0.000220
\n", + "
" + ], + "text/plain": [ + " Goal\n", + "SaTime1 0.055697\n", + "SaTime3 0.048387\n", + "SaTime2 0.048359\n", + "SaTime4 0.047626\n", + "absRoll 0.025873\n", + "climbRate 0.013537\n", + "Sgz 0.005171\n", + "curRoll 0.000220" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "elevators_correlations = compute_predictor_correlations(\n", + " data=df_elevators,\n", + " predictors=elevators_predictors,\n", + " imputed_variables=[elevators_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(elevators_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(elevators_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(elevators_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(elevators_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=elevators_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 14, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "a8796c1175244bfe8b0d41bed1cc6609", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/8 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
3absRoll0.00146301.562464e-0411.9552470.0013070
2curRoll0.00145801.509115e-0411.5470460.0013070
0climbRate0.00141401.066808e-048.1627160.0013070
1Sgz0.00135504.820981e-053.6887900.0013070
4SaTime10.00133102.440591e-051.8674260.0013070
7SaTime40.00131407.272511e-060.5564590.0013070
6SaTime30.0013070-6.777108e-08-0.0051860.0013070
5SaTime20.0013060-5.084813e-07-0.0389070.0013070
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "3 absRoll 0.001463 0 1.562464e-04 \n", + "2 curRoll 0.001458 0 1.509115e-04 \n", + "0 climbRate 0.001414 0 1.066808e-04 \n", + "1 Sgz 0.001355 0 4.820981e-05 \n", + "4 SaTime1 0.001331 0 2.440591e-05 \n", + "7 SaTime4 0.001314 0 7.272511e-06 \n", + "6 SaTime3 0.001307 0 -6.777108e-08 \n", + "5 SaTime2 0.001306 0 -5.084813e-07 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "3 11.955247 0.001307 0 \n", + "2 11.547046 0.001307 0 \n", + "0 8.162716 0.001307 0 \n", + "1 3.688790 0.001307 0 \n", + "4 1.867426 0.001307 0 \n", + "7 0.556459 0.001307 0 \n", + "6 -0.005186 0.001307 0 \n", + "5 -0.038907 0.001307 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis...\")\n", + "elevators_loo = leave_one_out_analysis(\n", + " data=df_elevators,\n", + " predictors=elevators_predictors,\n", + " imputed_variables=[elevators_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(elevators_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 15, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 9959\n", + "Receiver size: 6640\n", + "\n", + "Running autoimpute...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "d1202b9e6f5943fb9588f3ff0ef0bc42", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.00041733075698071036, + 0.0006945274770604267, + 0.0009126544374383521, + 0.0011107153879203403, + 0.0012569050309727105, + 0.0014289171149691683, + 0.0015593428510627136, + 0.0016715553625854499, + 0.0017485400744111554, + 0.0018047005450228843, + 0.0017793039233780928, + 0.0017537879897288806, + 0.001709326862951555, + 0.001633510137990435, + 0.0015007547013770807, + 0.0013647536706343202, + 0.0011642097218608235, + 0.0009397450424500616, + 0.0006514380374335112 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.00001016377262637028, + 0.000016322284753405822, + 0.000018987350280795512, + 0.00001816726623433957, + 0.000018542001529198758, + 0.000020704097797901034, + 0.000021673718592615825, + 0.000023117582879046623, + 0.000024665445964902992, + 0.00002570622785701564, + 0.00002791406542225051, + 0.000033108083242520206, + 0.00003834331900962756, + 0.00004268819946188325, + 0.00004613676606862346, + 0.00004734341041703108, + 0.00004705324688034628, + 0.000044173751641560596, + 0.000041323795815408205 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.00042523916093980705, + 0.0007002687973446931, + 0.0009127106131948651, + 0.0010828208840845696, + 0.0012231928812307582, + 0.0013408541000878575, + 0.0014416738271257018, + 0.0015249423539199505, + 0.0015901556236961128, + 0.0016398446421357926, + 0.001671536236653303, + 0.0016807631390262927, + 0.0016638363311461753, + 0.0016178087853864383, + 0.001536460577821025, + 0.0014138034451574141, + 0.0012402417531022125, + 0.0009999176374546677, + 0.0006625295568369386 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.000010084322371820636, + 0.0000159273685916936, + 0.000020316394108550417, + 0.00002502702016488793, + 0.00002952830318623912, + 0.00003181429293905948, + 0.00003323814954964788, + 0.00003455508726395601, + 0.00003729659692395035, + 0.0000396063558108858, + 0.000042004792443265195, + 0.0000435656599070157, + 0.000043505840286401174, + 0.000044223567391292146, + 0.00004437533549716239, + 0.00004081007347614774, + 0.00003288217809198435, + 0.000027547328921898425, + 0.000021850959006030572 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0002817784169756737, + 0.0005113577150345891, + 0.0007144241132768496, + 0.0008961487871232647, + 0.0010595183080782506, + 0.0012034494757487118, + 0.0013292688760591382, + 0.0014318493580148816, + 0.0015135825366891982, + 0.0015732220659344747, + 0.001607286011995681, + 0.001617206960286656, + 0.0015990604549444656, + 0.0015507169047136707, + 0.001469917186888279, + 0.0013516167592724619, + 0.0011785829817773212, + 0.0009424738375426881, + 0.0006030690360714073 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.00016461815990283907, + 0.00015311689125606818, + 0.00014182058836153944, + 0.00013078237364288925, + 0.00012007344546271649, + 0.00010979020189466915, + 0.0001000639677207584, + 0.00009107337655503052, + 0.00008305766235220195, + 0.0000763245944034747, + 0.00007123879016001055, + 0.00006816993335212892, + 0.00006739412051633259, + 0.00006898875433192543, + 0.0000727982283131681, + 0.00007850075760122417, + 0.00008571936140429814, + 0.00009410580201198965, + 0.00010337624534796863 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0025563195070588732, + 0.0025649158260364416, + 0.0025735121450140086, + 0.0025821084639915766, + 0.0025907047829691445, + 0.0025993011019467115, + 0.00260789742092428, + 0.002616493739901848, + 0.0026250900588794153, + 0.002633686377856983, + 0.0026422826968345507, + 0.0026508790158121186, + 0.0026594753347896857, + 0.002668071653767254, + 0.002676667972744822, + 0.0026852642917223894, + 0.0026938606106999574, + 0.002702456929677525, + 0.002711053248655093 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.012748900086031252, + 0.01846694746217978, + 0.02083981882861257, + 0.02216094526331568, + 0.021535820349572767, + 0.020391680612038707, + 0.019511224717398017, + 0.01767458689932801, + 0.017130388858639684, + 0.015366509392069976, + 0.012751788597809451, + 0.010685870542255928, + 0.009577198397340676, + 0.008226655534004472, + 0.006859804131544297, + 0.005400338302880356, + 0.0047259279217197695, + 0.0032804109515719234, + 0.002846915284857999 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.012602380012443976, + 0.016513906298007766, + 0.018324844300578954, + 0.019283949361179616, + 0.01913778648904192, + 0.01832900282100227, + 0.018181329483216564, + 0.017506212654085765, + 0.017624749494014276, + 0.016933351144075036, + 0.01615656918312625, + 0.015366839451844794, + 0.014958599544623951, + 0.014316459444683854, + 0.013600495835174111, + 0.012506342483222849, + 0.01161429740353466, + 0.009405304824591973, + 0.00668313134903145 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0013211589013804564, + "y1": 0.0013211589013804564 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0012825579129655041, + "y1": 0.0012825579129655041 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0011807647256014558, + "y1": 0.0011807647256014558 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0026336863778569828, + "y1": 0.0026336863778569828 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.015212923767235787, + "y1": 0.015212923767235787 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Elevators Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "elevators_donor, elevators_receiver = train_test_split(\n", + " df_elevators, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "elevators_receiver_no_target = elevators_receiver.drop(columns=[elevators_target])\n", + "\n", + "print(f\"Donor size: {len(elevators_donor)}\")\n", + "print(f\"Receiver size: {len(elevators_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute...\")\n", + "elevators_result = autoimpute(\n", + " donor_data=elevators_donor,\n", + " receiver_data=elevators_receiver_no_target.copy(),\n", + " predictors=elevators_predictors,\n", + " imputed_variables=[elevators_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=elevators_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Elevators Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 16, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Autoimpute CV Results (Zoomed In Without MDN) ---\n" + ] + }, + { + "data": { + "application/vnd.plotly.v1+json": { + "config": { + "plotlyServerURL": "https://plot.ly" + }, + "data": [ + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.000007279832919092497, + 0.00002946086781168022, + 0.000017680015605553013, + 0.00003260064726269813, + 0.00003557650694753496, + 0.000019613045828998836, + 0.00001814932280261359, + 0.000028694976305096084, + 0.000046368533212824535, + 0.000024293989296681853, + 0.00003541115170254013, + 0.00002710084542648461, + 0.000031513500876534857, + 0.000020319534850442018, + 0.000021683408229124396, + 0.000027385567802449857, + 0.00002121114757429098, + 0.000024668014215883887, + 0.00001862200308553563 + ] + }, + "hovertemplate": "Method=QRF
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.00041733075698071036, + 0.0006945274770604267, + 0.0009126544374383521, + 0.0011107153879203403, + 0.0012569050309727105, + 0.0014289171149691683, + 0.0015593428510627136, + 0.0016715553625854499, + 0.0017485400744111554, + 0.0018047005450228843, + 0.0017793039233780928, + 0.0017537879897288806, + 0.001709326862951555, + 0.001633510137990435, + 0.0015007547013770807, + 0.0013647536706343202, + 0.0011642097218608235, + 0.0009397450424500616, + 0.0006514380374335112 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.00001016377262637028, + 0.000016322284753405822, + 0.000018987350280795512, + 0.00001816726623433957, + 0.000018542001529198758, + 0.000020704097797901034, + 0.000021673718592615825, + 0.000023117582879046623, + 0.000024665445964902992, + 0.00002570622785701564, + 0.00002791406542225051, + 0.000033108083242520206, + 0.00003834331900962756, + 0.00004268819946188325, + 0.00004613676606862346, + 0.00004734341041703108, + 0.00004705324688034628, + 0.000044173751641560596, + 0.000041323795815408205 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.00042523916093980705, + 0.0007002687973446931, + 0.0009127106131948651, + 0.0010828208840845696, + 0.0012231928812307582, + 0.0013408541000878575, + 0.0014416738271257018, + 0.0015249423539199505, + 0.0015901556236961128, + 0.0016398446421357926, + 0.001671536236653303, + 0.0016807631390262927, + 0.0016638363311461753, + 0.0016178087853864383, + 0.001536460577821025, + 0.0014138034451574141, + 0.0012402417531022125, + 0.0009999176374546677, + 0.0006625295568369386 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.000010084322371820636, + 0.0000159273685916936, + 0.000020316394108550417, + 0.00002502702016488793, + 0.00002952830318623912, + 0.00003181429293905948, + 0.00003323814954964788, + 0.00003455508726395601, + 0.00003729659692395035, + 0.0000396063558108858, + 0.000042004792443265195, + 0.0000435656599070157, + 0.000043505840286401174, + 0.000044223567391292146, + 0.00004437533549716239, + 0.00004081007347614774, + 0.00003288217809198435, + 0.000027547328921898425, + 0.000021850959006030572 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0002817784169756737, + 0.0005113577150345891, + 0.0007144241132768496, + 0.0008961487871232647, + 0.0010595183080782506, + 0.0012034494757487118, + 0.0013292688760591382, + 0.0014318493580148816, + 0.0015135825366891982, + 0.0015732220659344747, + 0.001607286011995681, + 0.001617206960286656, + 0.0015990604549444656, + 0.0015507169047136707, + 0.001469917186888279, + 0.0013516167592724619, + 0.0011785829817773212, + 0.0009424738375426881, + 0.0006030690360714073 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.00016461815990283907, + 0.00015311689125606818, + 0.00014182058836153944, + 0.00013078237364288925, + 0.00012007344546271649, + 0.00010979020189466915, + 0.0001000639677207584, + 0.00009107337655503052, + 0.00008305766235220195, + 0.0000763245944034747, + 0.00007123879016001055, + 0.00006816993335212892, + 0.00006739412051633259, + 0.00006898875433192543, + 0.0000727982283131681, + 0.00007850075760122417, + 0.00008571936140429814, + 0.00009410580201198965, + 0.00010337624534796863 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0025563195070588732, + 0.0025649158260364416, + 0.0025735121450140086, + 0.0025821084639915766, + 0.0025907047829691445, + 0.0025993011019467115, + 0.00260789742092428, + 0.002616493739901848, + 0.0026250900588794153, + 0.002633686377856983, + 0.0026422826968345507, + 0.0026508790158121186, + 0.0026594753347896857, + 0.002668071653767254, + 0.002676667972744822, + 0.0026852642917223894, + 0.0026938606106999574, + 0.002702456929677525, + 0.002711053248655093 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0013211589013804564, + "y1": 0.0013211589013804564 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0012825579129655041, + "y1": 0.0012825579129655041 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0011807647256014558, + "y1": 0.0011807647256014558 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0026336863778569828, + "y1": 0.0026336863778569828 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Elevators Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "elevators_result.cv_results.pop('MDN', None)\n", + "\n", + "print(\"\\n--- Autoimpute CV Results (Zoomed In Without MDN) ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=elevators_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Elevators Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 17, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (Elevators) ---\n", + "QuantReg: Wasserstein Distance = 0.0015\n", + "QRF: Wasserstein Distance = 0.0003\n", + "OLS: Wasserstein Distance = 0.0014\n", + "Matching: Wasserstein Distance = 0.0002\n", + "MDN: Wasserstein Distance = 0.0200\n" + ] + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (Elevators) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "elevators_wasserstein = {}\n", + "\n", + "distribution_comparison_elevators = compare_distributions(\n", + " donor_data=elevators_receiver, # Ground truth test split\n", + " receiver_data=elevators_result.receiver_data, # Imputed values\n", + " imputed_variables=[elevators_target],\n", + ")\n", + "best_method_name = elevators_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "elevators_wasserstein[best_method_name] = distribution_comparison_elevators[\n", + " distribution_comparison_elevators['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {elevators_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in elevators_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = elevators_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [elevators_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=elevators_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[elevators_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " elevators_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {elevators_wasserstein.get(method_name, 'N/A'):.4f}\")" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 18, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "a42e0008daff49d3a8df67366598ebb1", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/8 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
cityarearoomsbathroomparking_spacesflooranimalfurniturehoa_(BRL)rent_amount_(BRL)property_tax_(BRL)fire_insurance_(BRL)total_(BRL)
0Sao Paulo702117aceptfurnished20653300211425618
1Sao Paulo32044020aceptnot furnished120049601750637973
2Porto Alegre801116aceptnot furnished100028000413841
3Porto Alegre512102aceptnot furnished270111222171421
4Sao Paulo251101not aceptnot furnished08002511836
\n", + "" + ], + "text/plain": [ + " city area rooms bathroom parking_spaces floor animal \\\n", + "0 Sao Paulo 70 2 1 1 7 acept \n", + "1 Sao Paulo 320 4 4 0 20 acept \n", + "2 Porto Alegre 80 1 1 1 6 acept \n", + "3 Porto Alegre 51 2 1 0 2 acept \n", + "4 Sao Paulo 25 1 1 0 1 not acept \n", + "\n", + " furniture hoa_(BRL) rent_amount_(BRL) property_tax_(BRL) \\\n", + "0 furnished 2065 3300 211 \n", + "1 not furnished 1200 4960 1750 \n", + "2 not furnished 1000 2800 0 \n", + "3 not furnished 270 1112 22 \n", + "4 not furnished 0 800 25 \n", + "\n", + " fire_insurance_(BRL) total_(BRL) \n", + "0 42 5618 \n", + "1 63 7973 \n", + "2 41 3841 \n", + "3 17 1421 \n", + "4 11 836 " + ] + }, + "execution_count": 19, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_brazilian = load_dataset(\"brazilian_houses\")\n", + "print(f\"\\nShape: {df_brazilian.shape}\")\n", + "print(f\"\\nColumn names:\")\n", + "for i, col in enumerate(df_brazilian.columns):\n", + " print(f\" {i+1}. {col}\")\n", + "print(f\"\\nData types:\")\n", + "print(df_brazilian.dtypes)\n", + "print(f\"\\nFirst few rows:\")\n", + "df_brazilian.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 20, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: rent_amount_(BRL)\n", + "Predictors (5): ['rooms', 'floor', 'hoa_(BRL)', 'property_tax_(BRL)', 'fire_insurance_(BRL)']\n" + ] + } + ], + "source": [ + "brazilian_target = \"rent_amount_(BRL)\" # Target variable to impute\n", + "brazilian_predictors = [\"rooms\", \"floor\", \"hoa_(BRL)\", \"property_tax_(BRL)\", \"fire_insurance_(BRL)\"]\n", + "\n", + "print(f\"Target variable: {brazilian_target}\")\n", + "print(f\"Predictors ({len(brazilian_predictors)}): {brazilian_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 21, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
roomsfloorhoa_(BRL)property_tax_(BRL)fire_insurance_(BRL)
rooms1.000000-0.0786870.0071390.0752520.565148
floor-0.0786871.0000000.0198720.0126260.013652
hoa_(BRL)0.0071390.0198721.0000000.0076270.029535
property_tax_(BRL)0.0752520.0126260.0076271.0000000.105661
fire_insurance_(BRL)0.5651480.0136520.0295350.1056611.000000
\n", + "
" + ], + "text/plain": [ + " rooms floor hoa_(BRL) property_tax_(BRL) \\\n", + "rooms 1.000000 -0.078687 0.007139 0.075252 \n", + "floor -0.078687 1.000000 0.019872 0.012626 \n", + "hoa_(BRL) 0.007139 0.019872 1.000000 0.007627 \n", + "property_tax_(BRL) 0.075252 0.012626 0.007627 1.000000 \n", + "fire_insurance_(BRL) 0.565148 0.013652 0.029535 0.105661 \n", + "\n", + " fire_insurance_(BRL) \n", + "rooms 0.565148 \n", + "floor 0.013652 \n", + "hoa_(BRL) 0.029535 \n", + "property_tax_(BRL) 0.105661 \n", + "fire_insurance_(BRL) 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
roomsfloorhoa_(BRL)property_tax_(BRL)fire_insurance_(BRL)
rooms1.000000-0.1173450.2061820.5955160.613788
floor-0.1173451.0000000.6182500.048770-0.004572
hoa_(BRL)0.2061820.6182501.0000000.3925370.293228
property_tax_(BRL)0.5955160.0487700.3925371.0000000.656049
fire_insurance_(BRL)0.613788-0.0045720.2932280.6560491.000000
\n", + "
" + ], + "text/plain": [ + " rooms floor hoa_(BRL) property_tax_(BRL) \\\n", + "rooms 1.000000 -0.117345 0.206182 0.595516 \n", + "floor -0.117345 1.000000 0.618250 0.048770 \n", + "hoa_(BRL) 0.206182 0.618250 1.000000 0.392537 \n", + "property_tax_(BRL) 0.595516 0.048770 0.392537 1.000000 \n", + "fire_insurance_(BRL) 0.613788 -0.004572 0.293228 0.656049 \n", + "\n", + " fire_insurance_(BRL) \n", + "rooms 0.613788 \n", + "floor -0.004572 \n", + "hoa_(BRL) 0.293228 \n", + "property_tax_(BRL) 0.656049 \n", + "fire_insurance_(BRL) 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
roomsfloorhoa_(BRL)property_tax_(BRL)fire_insurance_(BRL)
rooms1.0000000.0196680.1029920.1185880.119550
floor0.0196681.0000000.1072080.0142030.039879
hoa_(BRL)0.1029920.1072081.0000000.0651450.066869
property_tax_(BRL)0.1185880.0142030.0651451.0000000.063229
fire_insurance_(BRL)0.1195500.0398790.0668690.0632291.000000
\n", + "
" + ], + "text/plain": [ + " rooms floor hoa_(BRL) property_tax_(BRL) \\\n", + "rooms 1.000000 0.019668 0.102992 0.118588 \n", + "floor 0.019668 1.000000 0.107208 0.014203 \n", + "hoa_(BRL) 0.102992 0.107208 1.000000 0.065145 \n", + "property_tax_(BRL) 0.118588 0.014203 0.065145 1.000000 \n", + "fire_insurance_(BRL) 0.119550 0.039879 0.066869 0.063229 \n", + "\n", + " fire_insurance_(BRL) \n", + "rooms 0.119550 \n", + "floor 0.039879 \n", + "hoa_(BRL) 0.066869 \n", + "property_tax_(BRL) 0.063229 \n", + "fire_insurance_(BRL) 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
rent_amount_(BRL)
fire_insurance_(BRL)0.412287
property_tax_(BRL)0.057195
hoa_(BRL)0.051921
rooms0.031211
floor0.009949
\n", + "
" + ], + "text/plain": [ + " rent_amount_(BRL)\n", + "fire_insurance_(BRL) 0.412287\n", + "property_tax_(BRL) 0.057195\n", + "hoa_(BRL) 0.051921\n", + "rooms 0.031211\n", + "floor 0.009949" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "brazilian_correlations = compute_predictor_correlations(\n", + " data=df_brazilian,\n", + " predictors=brazilian_predictors,\n", + " imputed_variables=[brazilian_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(brazilian_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(brazilian_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(brazilian_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(brazilian_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=brazilian_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 22, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "27335e43812e468c9dd620f897266fdc", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/5 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
4fire_insurance_(BRL)662.9490940611.6006441191.07908951.3484490
1floor56.87479405.52634510.76243751.3484490
2hoa_(BRL)53.95599702.6075475.07814251.3484490
3property_tax_(BRL)52.57312401.2246742.38502751.3484490
0rooms52.34173400.9932851.93440151.3484490
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "4 fire_insurance_(BRL) 662.949094 0 611.600644 \n", + "1 floor 56.874794 0 5.526345 \n", + "2 hoa_(BRL) 53.955997 0 2.607547 \n", + "3 property_tax_(BRL) 52.573124 0 1.224674 \n", + "0 rooms 52.341734 0 0.993285 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "4 1191.079089 51.348449 0 \n", + "1 10.762437 51.348449 0 \n", + "2 5.078142 51.348449 0 \n", + "3 2.385027 51.348449 0 \n", + "0 1.934401 51.348449 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis...\")\n", + "brazilian_loo = leave_one_out_analysis(\n", + " data=df_brazilian,\n", + " predictors=brazilian_predictors,\n", + " imputed_variables=[brazilian_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(brazilian_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 23, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 6415\n", + "Receiver size: 4277\n", + "\n", + "Running autoimpute...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "f796139cda7e4e28a2eab0e7474d88d9", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 27.434123148869837, + 35.71683554169915, + 43.922470771628994, + 49.21275136399065, + 55.548480124707716, + 60.572860483242394, + 65.50799688230708, + 66.43289166017146, + 67.35795011691349, + 65.8282151208106, + 64.32813717848792, + 62.38007794232267, + 60.019088074824616, + 56.780280592361656, + 52.44731098986749, + 47.88885424785658, + 42.467918939984415, + 37.35066250974279, + 29.9317614964926 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 9.557234604861549, + 8.956439155330603, + 8.289368354662047, + 7.596266973236795, + 6.9520128338534635, + 6.296534342783538, + 5.6211587352247445, + 4.908122927624412, + 3.7955282448487293, + 2.818644336351974, + 2.09174379627377, + 1.8420185929381039, + 2.05294500583899, + 2.2709424329333614, + 2.6048666108839944, + 2.8950856016426756, + 3.289546046612837, + 3.475716713562988, + 3.1826312874153015 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 60.21841618347967, + 89.20306359511531, + 109.6712176243354, + 124.46568009905874, + 134.78429758257494, + 141.76553013766224, + 146.21444277019924, + 148.99821816017044, + 151.32972095844445, + 153.51725530867773, + 154.60351178771742, + 154.74132339224963, + 153.35356004060458, + 149.65104502519253, + 142.39016058555669, + 131.17725926043133, + 114.78010041972375, + 91.63331439897473, + 59.110454376877534 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 8.15670690198403, + 7.809585963580716, + 5.578023744610601, + 5.913426158297302, + 8.782705554389233, + 2.656850537091216, + 2.479605603420727, + 10.004387902935983, + 15.30913095948508, + 15.387617206801348, + 10.2752171682886, + 8.783492202987343, + 7.602765808678525, + 5.96440876713662, + 5.116530437829269, + 15.632955439876456, + 31.718604969178777, + 3.770810507561032, + 3.564229901764414 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 43.39814190414394, + 69.36283126903236, + 85.77857204469834, + 102.36882753066385, + 114.15020625795276, + 122.93034306885575, + 133.79421716721043, + 152.16683197699382, + 167.39144465711925, + 162.35099337948319, + 147.43137042156485, + 131.3510126289026, + 118.20775772834531, + 104.96705980967423, + 91.40677696892523, + 85.65733745986346, + 79.54944934430779, + 48.28930320097427, + 33.231650551214386 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 20.787085449742314, + 18.133934780109254, + 15.596770860589444, + 13.242429289987362, + 11.186939137986156, + 9.623735377066708, + 8.818614374872773, + 8.977897188747919, + 10.055859879933255, + 11.803431682018678, + 13.971566933362238, + 16.394247992172986, + 18.97422075971559, + 21.655339784960336, + 24.404291231663677, + 27.200516932435963, + 30.030814404836313, + 32.88638788006087, + 35.76118291292973 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 354.65874512860483, + 359.5553702260328, + 364.45199532346066, + 369.34862042088855, + 374.24524551831644, + 379.1418706157443, + 384.03849571317227, + 388.93512081060015, + 393.831745908028, + 398.728371005456, + 403.6249961028839, + 408.5216212003118, + 413.41824629773964, + 418.3148713951676, + 423.2114964925954, + 428.1081215900234, + 433.0047466874513, + 437.9013717848792, + 442.7979968823071 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 17.82708018826267, + 25.705230261351847, + 19.507534048055877, + 14.297240685602839, + 33.842332189775505, + 66.59788185374984, + 105.56188466433146, + 150.7109411202626, + 200.2093336892162, + 254.519039676545, + 312.86215715520075, + 375.71540655326277, + 442.0045411281469, + 512.7219798864934, + 587.260642342192, + 667.7304038811378, + 753.0074651554911, + 845.6548833756536, + 947.8124427082157 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 211.95363543779845, + 406.6583397522585, + 588.5803247627703, + 760.992829991829, + 925.0212772214775, + 1080.747056574654, + 1230.5342890000138, + 1373.8658721081802, + 1511.7503385087427, + 1643.7695372411265, + 1770.2148889244006, + 1890.0146683348182, + 2004.0096345906543, + 2110.7926275320788, + 2209.449751827035, + 2297.591575550011, + 2372.695563690352, + 2427.2053368165393, + 2443.2381622618914 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 52.16466669401485, + "y1": 52.16466669401485 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 126.92676693194979, + "y1": 126.92676693194979 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 104.9360067036803, + "y1": 104.9360067036803 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 398.72837100545604, + "y1": 398.72837100545604 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1539.9518794803491, + "y1": 1539.9518794803491 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Brazilian Houses Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "brazilian_donor, brazilian_receiver = train_test_split(\n", + " df_brazilian, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "brazilian_receiver_no_target = brazilian_receiver.drop(columns=[brazilian_target])\n", + "\n", + "print(f\"Donor size: {len(brazilian_donor)}\")\n", + "print(f\"Receiver size: {len(brazilian_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute...\")\n", + "brazilian_result = autoimpute(\n", + " donor_data=brazilian_donor,\n", + " receiver_data=brazilian_receiver_no_target.copy(),\n", + " predictors=brazilian_predictors,\n", + " imputed_variables=[brazilian_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=brazilian_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Brazilian Houses Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 24, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Autoimpute CV Results (Zoomed In Without MDN) ---\n" + ] + }, + { + "data": { + "application/vnd.plotly.v1+json": { + "config": { + "plotlyServerURL": "https://plot.ly" + }, + "data": [ + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 3.1072974576558354, + 3.173472785580769, + 2.845580589626428, + 1.7549878759386435, + 4.177878882735426, + 3.347338117083344, + 3.3550935697327104, + 3.649098163471762, + 4.019663168790427, + 4.78874671805982, + 6.2170904100295425, + 4.460696816140001, + 7.5800773055275545, + 7.8227182830889435, + 7.401020283305162, + 10.023676321476369, + 9.723242741104022, + 10.253533450814857, + 11.13805573184919 + ] + }, + "hovertemplate": "Method=QRF
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 27.434123148869837, + 35.71683554169915, + 43.922470771628994, + 49.21275136399065, + 55.548480124707716, + 60.572860483242394, + 65.50799688230708, + 66.43289166017146, + 67.35795011691349, + 65.8282151208106, + 64.32813717848792, + 62.38007794232267, + 60.019088074824616, + 56.780280592361656, + 52.44731098986749, + 47.88885424785658, + 42.467918939984415, + 37.35066250974279, + 29.9317614964926 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 9.557234604861549, + 8.956439155330603, + 8.289368354662047, + 7.596266973236795, + 6.9520128338534635, + 6.296534342783538, + 5.6211587352247445, + 4.908122927624412, + 3.7955282448487293, + 2.818644336351974, + 2.09174379627377, + 1.8420185929381039, + 2.05294500583899, + 2.2709424329333614, + 2.6048666108839944, + 2.8950856016426756, + 3.289546046612837, + 3.475716713562988, + 3.1826312874153015 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 60.21841618347967, + 89.20306359511531, + 109.6712176243354, + 124.46568009905874, + 134.78429758257494, + 141.76553013766224, + 146.21444277019924, + 148.99821816017044, + 151.32972095844445, + 153.51725530867773, + 154.60351178771742, + 154.74132339224963, + 153.35356004060458, + 149.65104502519253, + 142.39016058555669, + 131.17725926043133, + 114.78010041972375, + 91.63331439897473, + 59.110454376877534 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 8.15670690198403, + 7.809585963580716, + 5.578023744610601, + 5.913426158297302, + 8.782705554389233, + 2.656850537091216, + 2.479605603420727, + 10.004387902935983, + 15.30913095948508, + 15.387617206801348, + 10.2752171682886, + 8.783492202987343, + 7.602765808678525, + 5.96440876713662, + 5.116530437829269, + 15.632955439876456, + 31.718604969178777, + 3.770810507561032, + 3.564229901764414 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 43.39814190414394, + 69.36283126903236, + 85.77857204469834, + 102.36882753066385, + 114.15020625795276, + 122.93034306885575, + 133.79421716721043, + 152.16683197699382, + 167.39144465711925, + 162.35099337948319, + 147.43137042156485, + 131.3510126289026, + 118.20775772834531, + 104.96705980967423, + 91.40677696892523, + 85.65733745986346, + 79.54944934430779, + 48.28930320097427, + 33.231650551214386 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 20.787085449742314, + 18.133934780109254, + 15.596770860589444, + 13.242429289987362, + 11.186939137986156, + 9.623735377066708, + 8.818614374872773, + 8.977897188747919, + 10.055859879933255, + 11.803431682018678, + 13.971566933362238, + 16.394247992172986, + 18.97422075971559, + 21.655339784960336, + 24.404291231663677, + 27.200516932435963, + 30.030814404836313, + 32.88638788006087, + 35.76118291292973 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 354.65874512860483, + 359.5553702260328, + 364.45199532346066, + 369.34862042088855, + 374.24524551831644, + 379.1418706157443, + 384.03849571317227, + 388.93512081060015, + 393.831745908028, + 398.728371005456, + 403.6249961028839, + 408.5216212003118, + 413.41824629773964, + 418.3148713951676, + 423.2114964925954, + 428.1081215900234, + 433.0047466874513, + 437.9013717848792, + 442.7979968823071 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 52.16466669401485, + "y1": 52.16466669401485 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 126.92676693194979, + "y1": 126.92676693194979 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 104.9360067036803, + "y1": 104.9360067036803 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 398.72837100545604, + "y1": 398.72837100545604 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Brazilian Houses Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "brazilian_result.cv_results.pop('MDN', None)\n", + "\n", + "print(\"\\n--- Autoimpute CV Results (Zoomed In Without MDN) ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=brazilian_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Brazilian Houses Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 25, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (Brazilian Houses) ---\n", + "QRF: Wasserstein Distance = 27.1747\n", + "OLS: Wasserstein Distance = 99.3468\n", + "QuantReg: Wasserstein Distance = 112.3643\n", + "Matching: Wasserstein Distance = 120.4604\n", + "MDN: Wasserstein Distance = 3378.6172\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
VariableMetricDistance
0rent_amount_(BRL)wasserstein_distance27.174655
\n", + "
" + ], + "text/plain": [ + " Variable Metric Distance\n", + "0 rent_amount_(BRL) wasserstein_distance 27.174655" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (Brazilian Houses) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "brazilian_wasserstein = {}\n", + "\n", + "distribution_comparison_brazilian = compare_distributions(\n", + " donor_data=brazilian_receiver, # Ground truth test split\n", + " receiver_data=brazilian_result.receiver_data, # Imputed values\n", + " imputed_variables=[brazilian_target],\n", + ")\n", + "best_method_name = brazilian_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "brazilian_wasserstein[best_method_name] = distribution_comparison_brazilian[\n", + " distribution_comparison_brazilian['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {brazilian_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in brazilian_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = brazilian_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [brazilian_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=brazilian_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[brazilian_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " brazilian_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {brazilian_wasserstein.get(method_name, 'N/A'):.4f}\")\n", + " \n", + "# Display full comparison for best method\n", + "display(distribution_comparison_brazilian)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 26, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "03a9802aea964ed0965c0e243041d378", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/5 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
timedeltan_tokens_titlen_tokens_contentn_unique_tokensn_non_stop_wordsn_non_stop_unique_tokensnum_hrefsnum_self_hrefsnum_imgsnum_videosaverage_token_lengthnum_keywordsdata_channel_is_lifestyledata_channel_is_entertainmentdata_channel_is_busdata_channel_is_socmeddata_channel_is_techdata_channel_is_worldkw_min_minkw_max_minkw_avg_minkw_min_maxkw_max_maxkw_avg_maxkw_min_avgkw_max_avgkw_avg_avgself_reference_min_sharesself_reference_max_sharesself_reference_avg_sharessweekday_is_mondayweekday_is_tuesdayweekday_is_wednesdayweekday_is_thursdayweekday_is_fridayweekday_is_saturdayweekday_is_sundayis_weekendLDA_00LDA_01LDA_02LDA_03LDA_04global_subjectivityglobal_sentiment_polarityglobal_rate_positive_wordsglobal_rate_negative_wordsrate_positive_wordsrate_negative_wordsavg_positive_polaritymin_positive_polaritymax_positive_polarityavg_negative_polaritymin_negative_polaritymax_negative_polaritytitle_subjectivitytitle_sentiment_polarityabs_title_subjectivityabs_title_sentiment_polarityshares
0731.012219.00.6635941.00.8153854.02104.68036550100000.00.00.00.00.00.00.00.00.0496.0496.0496.000000100000000.5003310.3782790.0400050.0412630.0401230.5216170.0925620.0456620.0136990.7692310.2307690.3786360.1000000.7-0.350000-0.600-0.2000000.500000-0.1875000.0000000.187500593.0
1731.09255.00.6047431.00.7919463.01104.91372540010000.00.00.00.00.00.00.00.00.00.00.00.000000100000000.7997560.0500470.0500960.0501010.0500010.3412460.1489480.0431370.0156860.7333330.2666670.2869150.0333330.7-0.118750-0.125-0.1000000.0000000.0000000.5000000.000000711.0
2731.09211.00.5751301.00.6638663.01104.39336560010000.00.00.00.00.00.00.00.00.0918.0918.0918.000000100000000.2177920.0333340.0333510.0333340.6821880.7022220.3233330.0568720.0094790.8571430.1428570.4958330.1000001.0-0.466667-0.800-0.1333330.0000000.0000000.5000000.0000001500.0
3731.09531.00.5037881.00.6656359.00104.40489670100000.00.00.00.00.00.00.00.00.00.00.00.000000100000000.0285730.4193000.4946510.0289050.0285720.4298500.1007050.0414310.0207160.6666670.3333330.3859650.1363640.8-0.369697-0.600-0.1666670.0000000.0000000.5000000.0000001200.0
4731.0131072.00.4156461.00.54089019.0192004.68283670000100.00.00.00.00.00.00.00.00.0545.016000.03151.157895100000000.0286330.0287940.0285750.0285720.8854270.5135020.2810030.0746270.0121270.8602150.1397850.4111270.0333331.0-0.220192-0.500-0.0500000.4545450.1363640.0454550.136364505.0
\n", + "" + ], + "text/plain": [ + " timedelta n_tokens_title n_tokens_content n_unique_tokens \\\n", + "0 731.0 12 219.0 0.663594 \n", + "1 731.0 9 255.0 0.604743 \n", + "2 731.0 9 211.0 0.575130 \n", + "3 731.0 9 531.0 0.503788 \n", + "4 731.0 13 1072.0 0.415646 \n", + "\n", + " n_non_stop_words n_non_stop_unique_tokens num_hrefs num_self_hrefs \\\n", + "0 1.0 0.815385 4.0 2 \n", + "1 1.0 0.791946 3.0 1 \n", + "2 1.0 0.663866 3.0 1 \n", + "3 1.0 0.665635 9.0 0 \n", + "4 1.0 0.540890 19.0 19 \n", + "\n", + " num_imgs num_videos average_token_length num_keywords \\\n", + "0 1 0 4.680365 5 \n", + "1 1 0 4.913725 4 \n", + "2 1 0 4.393365 6 \n", + "3 1 0 4.404896 7 \n", + "4 20 0 4.682836 7 \n", + "\n", + " data_channel_is_lifestyle data_channel_is_entertainment \\\n", + "0 0 1 \n", + "1 0 0 \n", + "2 0 0 \n", + "3 0 1 \n", + "4 0 0 \n", + "\n", + " data_channel_is_bus data_channel_is_socmed data_channel_is_tech \\\n", + "0 0 0 0 \n", + "1 1 0 0 \n", + "2 1 0 0 \n", + "3 0 0 0 \n", + "4 0 0 1 \n", + "\n", + " data_channel_is_world kw_min_min kw_max_min kw_avg_min kw_min_max \\\n", + "0 0 0.0 0.0 0.0 0.0 \n", + "1 0 0.0 0.0 0.0 0.0 \n", + "2 0 0.0 0.0 0.0 0.0 \n", + "3 0 0.0 0.0 0.0 0.0 \n", + "4 0 0.0 0.0 0.0 0.0 \n", + "\n", + " kw_max_max kw_avg_max kw_min_avg kw_max_avg kw_avg_avg \\\n", + "0 0.0 0.0 0.0 0.0 0.0 \n", + "1 0.0 0.0 0.0 0.0 0.0 \n", + "2 0.0 0.0 0.0 0.0 0.0 \n", + "3 0.0 0.0 0.0 0.0 0.0 \n", + "4 0.0 0.0 0.0 0.0 0.0 \n", + "\n", + " self_reference_min_shares self_reference_max_shares \\\n", + "0 496.0 496.0 \n", + "1 0.0 0.0 \n", + "2 918.0 918.0 \n", + "3 0.0 0.0 \n", + "4 545.0 16000.0 \n", + "\n", + " self_reference_avg_sharess weekday_is_monday weekday_is_tuesday \\\n", + "0 496.000000 1 0 \n", + "1 0.000000 1 0 \n", + "2 918.000000 1 0 \n", + "3 0.000000 1 0 \n", + "4 3151.157895 1 0 \n", + "\n", + " weekday_is_wednesday weekday_is_thursday weekday_is_friday \\\n", + "0 0 0 0 \n", + "1 0 0 0 \n", + "2 0 0 0 \n", + "3 0 0 0 \n", + "4 0 0 0 \n", + "\n", + " weekday_is_saturday weekday_is_sunday is_weekend LDA_00 LDA_01 \\\n", + "0 0 0 0 0.500331 0.378279 \n", + "1 0 0 0 0.799756 0.050047 \n", + "2 0 0 0 0.217792 0.033334 \n", + "3 0 0 0 0.028573 0.419300 \n", + "4 0 0 0 0.028633 0.028794 \n", + "\n", + " LDA_02 LDA_03 LDA_04 global_subjectivity \\\n", + "0 0.040005 0.041263 0.040123 0.521617 \n", + "1 0.050096 0.050101 0.050001 0.341246 \n", + "2 0.033351 0.033334 0.682188 0.702222 \n", + "3 0.494651 0.028905 0.028572 0.429850 \n", + "4 0.028575 0.028572 0.885427 0.513502 \n", + "\n", + " global_sentiment_polarity global_rate_positive_words \\\n", + "0 0.092562 0.045662 \n", + "1 0.148948 0.043137 \n", + "2 0.323333 0.056872 \n", + "3 0.100705 0.041431 \n", + "4 0.281003 0.074627 \n", + "\n", + " global_rate_negative_words rate_positive_words rate_negative_words \\\n", + "0 0.013699 0.769231 0.230769 \n", + "1 0.015686 0.733333 0.266667 \n", + "2 0.009479 0.857143 0.142857 \n", + "3 0.020716 0.666667 0.333333 \n", + "4 0.012127 0.860215 0.139785 \n", + "\n", + " avg_positive_polarity min_positive_polarity max_positive_polarity \\\n", + "0 0.378636 0.100000 0.7 \n", + "1 0.286915 0.033333 0.7 \n", + "2 0.495833 0.100000 1.0 \n", + "3 0.385965 0.136364 0.8 \n", + "4 0.411127 0.033333 1.0 \n", + "\n", + " avg_negative_polarity min_negative_polarity max_negative_polarity \\\n", + "0 -0.350000 -0.600 -0.200000 \n", + "1 -0.118750 -0.125 -0.100000 \n", + "2 -0.466667 -0.800 -0.133333 \n", + "3 -0.369697 -0.600 -0.166667 \n", + "4 -0.220192 -0.500 -0.050000 \n", + "\n", + " title_subjectivity title_sentiment_polarity abs_title_subjectivity \\\n", + "0 0.500000 -0.187500 0.000000 \n", + "1 0.000000 0.000000 0.500000 \n", + "2 0.000000 0.000000 0.500000 \n", + "3 0.000000 0.000000 0.500000 \n", + "4 0.454545 0.136364 0.045455 \n", + "\n", + " abs_title_sentiment_polarity shares \n", + "0 0.187500 593.0 \n", + "1 0.000000 711.0 \n", + "2 0.000000 1500.0 \n", + "3 0.000000 1200.0 \n", + "4 0.136364 505.0 " + ] + }, + "execution_count": 27, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_news = load_dataset(\"onlinenewspopularity\")\n", + "print(f\"\\nShape: {df_news.shape}\")\n", + "print(f\"\\nColumn names (first 30 and last 10):\")\n", + "cols = list(df_news.columns)\n", + "for i, col in enumerate(cols[:30]):\n", + " print(f\" {i+1}. {col}\")\n", + "print(\" ...\")\n", + "for i, col in enumerate(cols[-10:]):\n", + " print(f\" {len(cols)-10+i+1}. {col}\")\n", + "print(f\"\\nData types (sample):\")\n", + "print(df_news.dtypes[:10])\n", + "print(f\"\\nFirst few rows:\")\n", + "df_news.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 28, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: shares\n", + "Predictors 9: ['n_tokens_title', 'n_tokens_content', 'num_hrefs', 'num_imgs', 'num_videos', 'global_subjectivity', 'global_sentiment_polarity', 'title_subjectivity', 'title_sentiment_polarity']\n" + ] + } + ], + "source": [ + "news_target = \"shares\" # Target variable to impute\n", + "news_predictors = [\"n_tokens_title\", \"n_tokens_content\", \"num_hrefs\", \"num_imgs\", \"num_videos\", \"global_subjectivity\", \"global_sentiment_polarity\", \"title_subjectivity\", \"title_sentiment_polarity\"]\n", + "\n", + "print(f\"Target variable: {news_target}\")\n", + "print(f\"Predictors {len(news_predictors)}: {news_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 29, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
n_tokens_titlen_tokens_contentnum_hrefsnum_imgsnum_videosglobal_subjectivityglobal_sentiment_polaritytitle_subjectivitytitle_sentiment_polarity
n_tokens_title1.0000000.018160-0.053496-0.0088580.051460-0.056804-0.0722260.0772450.000240
n_tokens_content0.0181601.0000000.4230650.3426000.1036990.1278790.0219370.0044840.023358
num_hrefs-0.0534960.4230651.0000000.3426330.1145180.2034640.0868590.0439500.039041
num_imgs-0.0088580.3426000.3426331.000000-0.0673360.0804680.0210820.0568150.046310
num_videos0.0514600.1036990.114518-0.0673361.0000000.082052-0.0284340.0610280.021980
global_subjectivity-0.0568040.1278790.2034640.0804680.0820521.0000000.3394360.1141230.034075
global_sentiment_polarity-0.0722260.0219370.0868590.021082-0.0284340.3394361.0000000.0236200.238266
title_subjectivity0.0772450.0044840.0439500.0568150.0610280.1141230.0236201.0000000.232130
title_sentiment_polarity0.0002400.0233580.0390410.0463100.0219800.0340750.2382660.2321301.000000
\n", + "
" + ], + "text/plain": [ + " n_tokens_title n_tokens_content num_hrefs \\\n", + "n_tokens_title 1.000000 0.018160 -0.053496 \n", + "n_tokens_content 0.018160 1.000000 0.423065 \n", + "num_hrefs -0.053496 0.423065 1.000000 \n", + "num_imgs -0.008858 0.342600 0.342633 \n", + "num_videos 0.051460 0.103699 0.114518 \n", + "global_subjectivity -0.056804 0.127879 0.203464 \n", + "global_sentiment_polarity -0.072226 0.021937 0.086859 \n", + "title_subjectivity 0.077245 0.004484 0.043950 \n", + "title_sentiment_polarity 0.000240 0.023358 0.039041 \n", + "\n", + " num_imgs num_videos global_subjectivity \\\n", + "n_tokens_title -0.008858 0.051460 -0.056804 \n", + "n_tokens_content 0.342600 0.103699 0.127879 \n", + "num_hrefs 0.342633 0.114518 0.203464 \n", + "num_imgs 1.000000 -0.067336 0.080468 \n", + "num_videos -0.067336 1.000000 0.082052 \n", + "global_subjectivity 0.080468 0.082052 1.000000 \n", + "global_sentiment_polarity 0.021082 -0.028434 0.339436 \n", + "title_subjectivity 0.056815 0.061028 0.114123 \n", + "title_sentiment_polarity 0.046310 0.021980 0.034075 \n", + "\n", + " global_sentiment_polarity title_subjectivity \\\n", + "n_tokens_title -0.072226 0.077245 \n", + "n_tokens_content 0.021937 0.004484 \n", + "num_hrefs 0.086859 0.043950 \n", + "num_imgs 0.021082 0.056815 \n", + "num_videos -0.028434 0.061028 \n", + "global_subjectivity 0.339436 0.114123 \n", + "global_sentiment_polarity 1.000000 0.023620 \n", + "title_subjectivity 0.023620 1.000000 \n", + "title_sentiment_polarity 0.238266 0.232130 \n", + "\n", + " title_sentiment_polarity \n", + "n_tokens_title 0.000240 \n", + "n_tokens_content 0.023358 \n", + "num_hrefs 0.039041 \n", + "num_imgs 0.046310 \n", + "num_videos 0.021980 \n", + "global_subjectivity 0.034075 \n", + "global_sentiment_polarity 0.238266 \n", + "title_subjectivity 0.232130 \n", + "title_sentiment_polarity 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
n_tokens_titlen_tokens_contentnum_hrefsnum_imgsnum_videosglobal_subjectivityglobal_sentiment_polaritytitle_subjectivitytitle_sentiment_polarity
n_tokens_title1.0000000.010237-0.061430-0.0126700.116492-0.047503-0.0772220.1086840.011174
n_tokens_content0.0102371.0000000.5061760.391022-0.0373410.0730550.029948-0.0104570.010359
num_hrefs-0.0614300.5061761.0000000.349176-0.0326170.1932060.1305990.0206700.026945
num_imgs-0.0126700.3910220.3491761.000000-0.1684680.1135520.0469200.0235200.030852
num_videos0.116492-0.037341-0.032617-0.1684681.0000000.072735-0.0477640.0597270.006032
global_subjectivity-0.0475030.0730550.1932060.1135520.0727351.0000000.3367130.1345250.050840
global_sentiment_polarity-0.0772220.0299480.1305990.046920-0.0477640.3367131.0000000.0255080.222153
title_subjectivity0.108684-0.0104570.0206700.0235200.0597270.1345250.0255081.0000000.351178
title_sentiment_polarity0.0111740.0103590.0269450.0308520.0060320.0508400.2221530.3511781.000000
\n", + "
" + ], + "text/plain": [ + " n_tokens_title n_tokens_content num_hrefs \\\n", + "n_tokens_title 1.000000 0.010237 -0.061430 \n", + "n_tokens_content 0.010237 1.000000 0.506176 \n", + "num_hrefs -0.061430 0.506176 1.000000 \n", + "num_imgs -0.012670 0.391022 0.349176 \n", + "num_videos 0.116492 -0.037341 -0.032617 \n", + "global_subjectivity -0.047503 0.073055 0.193206 \n", + "global_sentiment_polarity -0.077222 0.029948 0.130599 \n", + "title_subjectivity 0.108684 -0.010457 0.020670 \n", + "title_sentiment_polarity 0.011174 0.010359 0.026945 \n", + "\n", + " num_imgs num_videos global_subjectivity \\\n", + "n_tokens_title -0.012670 0.116492 -0.047503 \n", + "n_tokens_content 0.391022 -0.037341 0.073055 \n", + "num_hrefs 0.349176 -0.032617 0.193206 \n", + "num_imgs 1.000000 -0.168468 0.113552 \n", + "num_videos -0.168468 1.000000 0.072735 \n", + "global_subjectivity 0.113552 0.072735 1.000000 \n", + "global_sentiment_polarity 0.046920 -0.047764 0.336713 \n", + "title_subjectivity 0.023520 0.059727 0.134525 \n", + "title_sentiment_polarity 0.030852 0.006032 0.050840 \n", + "\n", + " global_sentiment_polarity title_subjectivity \\\n", + "n_tokens_title -0.077222 0.108684 \n", + "n_tokens_content 0.029948 -0.010457 \n", + "num_hrefs 0.130599 0.020670 \n", + "num_imgs 0.046920 0.023520 \n", + "num_videos -0.047764 0.059727 \n", + "global_subjectivity 0.336713 0.134525 \n", + "global_sentiment_polarity 1.000000 0.025508 \n", + "title_subjectivity 0.025508 1.000000 \n", + "title_sentiment_polarity 0.222153 0.351178 \n", + "\n", + " title_sentiment_polarity \n", + "n_tokens_title 0.011174 \n", + "n_tokens_content 0.010359 \n", + "num_hrefs 0.026945 \n", + "num_imgs 0.030852 \n", + "num_videos 0.006032 \n", + "global_subjectivity 0.050840 \n", + "global_sentiment_polarity 0.222153 \n", + "title_subjectivity 0.351178 \n", + "title_sentiment_polarity 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
n_tokens_titlen_tokens_contentnum_hrefsnum_imgsnum_videosglobal_subjectivityglobal_sentiment_polaritytitle_subjectivitytitle_sentiment_polarity
n_tokens_title1.0000000.0008620.0003540.0035050.0032290.0002540.0009170.0082020.007850
n_tokens_content0.0008621.0000000.0557660.0502160.0251890.0173130.0155280.0046970.005491
num_hrefs0.0003540.0557661.0000000.0494070.0221490.0302010.0260790.0031550.001383
num_imgs0.0035050.0502160.0494071.0000000.0449290.0209420.0168040.0062520.007037
num_videos0.0032290.0251890.0221490.0449291.0000000.0187910.0168000.0061230.006016
global_subjectivity0.0002540.0173130.0302010.0209420.0187911.0000000.0133510.0064500.004167
global_sentiment_polarity0.0009170.0155280.0260790.0168040.0168000.0133511.0000000.0044100.011626
title_subjectivity0.0082020.0046970.0031550.0062520.0061230.0064500.0044101.0000000.330122
title_sentiment_polarity0.0078500.0054910.0013830.0070370.0060160.0041670.0116260.3301221.000000
\n", + "
" + ], + "text/plain": [ + " n_tokens_title n_tokens_content num_hrefs \\\n", + "n_tokens_title 1.000000 0.000862 0.000354 \n", + "n_tokens_content 0.000862 1.000000 0.055766 \n", + "num_hrefs 0.000354 0.055766 1.000000 \n", + "num_imgs 0.003505 0.050216 0.049407 \n", + "num_videos 0.003229 0.025189 0.022149 \n", + "global_subjectivity 0.000254 0.017313 0.030201 \n", + "global_sentiment_polarity 0.000917 0.015528 0.026079 \n", + "title_subjectivity 0.008202 0.004697 0.003155 \n", + "title_sentiment_polarity 0.007850 0.005491 0.001383 \n", + "\n", + " num_imgs num_videos global_subjectivity \\\n", + "n_tokens_title 0.003505 0.003229 0.000254 \n", + "n_tokens_content 0.050216 0.025189 0.017313 \n", + "num_hrefs 0.049407 0.022149 0.030201 \n", + "num_imgs 1.000000 0.044929 0.020942 \n", + "num_videos 0.044929 1.000000 0.018791 \n", + "global_subjectivity 0.020942 0.018791 1.000000 \n", + "global_sentiment_polarity 0.016804 0.016800 0.013351 \n", + "title_subjectivity 0.006252 0.006123 0.006450 \n", + "title_sentiment_polarity 0.007037 0.006016 0.004167 \n", + "\n", + " global_sentiment_polarity title_subjectivity \\\n", + "n_tokens_title 0.000917 0.008202 \n", + "n_tokens_content 0.015528 0.004697 \n", + "num_hrefs 0.026079 0.003155 \n", + "num_imgs 0.016804 0.006252 \n", + "num_videos 0.016800 0.006123 \n", + "global_subjectivity 0.013351 0.006450 \n", + "global_sentiment_polarity 1.000000 0.004410 \n", + "title_subjectivity 0.004410 1.000000 \n", + "title_sentiment_polarity 0.011626 0.330122 \n", + "\n", + " title_sentiment_polarity \n", + "n_tokens_title 0.007850 \n", + "n_tokens_content 0.005491 \n", + "num_hrefs 0.001383 \n", + "num_imgs 0.007037 \n", + "num_videos 0.006016 \n", + "global_subjectivity 0.004167 \n", + "global_sentiment_polarity 0.011626 \n", + "title_subjectivity 0.330122 \n", + "title_sentiment_polarity 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
shares
num_imgs0.002237
global_sentiment_polarity0.000848
num_hrefs0.000824
title_subjectivity0.000505
n_tokens_title0.000396
title_sentiment_polarity0.000390
global_subjectivity0.000360
n_tokens_content0.000000
num_videos0.000000
\n", + "
" + ], + "text/plain": [ + " shares\n", + "num_imgs 0.002237\n", + "global_sentiment_polarity 0.000848\n", + "num_hrefs 0.000824\n", + "title_subjectivity 0.000505\n", + "n_tokens_title 0.000396\n", + "title_sentiment_polarity 0.000390\n", + "global_subjectivity 0.000360\n", + "n_tokens_content 0.000000\n", + "num_videos 0.000000" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "news_correlations = compute_predictor_correlations(\n", + " data=df_news,\n", + " predictors=news_predictors,\n", + " imputed_variables=[news_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(news_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(news_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(news_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(news_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=news_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 30, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis on top predictors...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "0da6e91ad8bb4b82ac700476832fbb15", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/9 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
5global_subjectivity1446.384028022.4727431.5782401423.9112860
4num_videos1445.642897021.7316111.5261911423.9112860
6global_sentiment_polarity1436.540498012.6292120.8869381423.9112860
8title_sentiment_polarity1427.68554103.7742550.2650631423.9112860
7title_subjectivity1423.4780420-0.433243-0.0304261423.9112860
0n_tokens_title1423.1408390-0.770446-0.0541081423.9112860
1n_tokens_content1419.1512110-4.760075-0.3342961423.9112860
3num_imgs1413.1874950-10.723790-0.7531221423.9112860
2num_hrefs1400.8356710-23.075615-1.6205801423.9112860
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "5 global_subjectivity 1446.384028 0 22.472743 \n", + "4 num_videos 1445.642897 0 21.731611 \n", + "6 global_sentiment_polarity 1436.540498 0 12.629212 \n", + "8 title_sentiment_polarity 1427.685541 0 3.774255 \n", + "7 title_subjectivity 1423.478042 0 -0.433243 \n", + "0 n_tokens_title 1423.140839 0 -0.770446 \n", + "1 n_tokens_content 1419.151211 0 -4.760075 \n", + "3 num_imgs 1413.187495 0 -10.723790 \n", + "2 num_hrefs 1400.835671 0 -23.075615 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "5 1.578240 1423.911286 0 \n", + "4 1.526191 1423.911286 0 \n", + "6 0.886938 1423.911286 0 \n", + "8 0.265063 1423.911286 0 \n", + "7 -0.030426 1423.911286 0 \n", + "0 -0.054108 1423.911286 0 \n", + "1 -0.334296 1423.911286 0 \n", + "3 -0.753122 1423.911286 0 \n", + "2 -1.620580 1423.911286 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis on top predictors...\")\n", + "news_loo = leave_one_out_analysis(\n", + " data=df_news,\n", + " predictors=news_predictors,\n", + " imputed_variables=[news_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(news_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 31, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 23786\n", + "Receiver size: 15858\n", + "\n", + "Running autoimpute with top predictors...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "076836bfd28f43db96c909103a354b45", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 223.8630796172769, + 458.84026692638434, + 579.6705542907807, + 781.7957689873281, + 965.9503445067965, + 1110.6244506911476, + 1263.9309570621926, + 1396.3557107328745, + 1567.414674054377, + 1666.0733698079766, + 1774.2381807111894, + 1841.788580990753, + 1877.6493608909611, + 1899.9221622620603, + 1903.3988327084717, + 1856.8536603963116, + 1739.2884707393005, + 1574.7065993779393, + 1304.629534601914 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 60.66262042015628, + 92.40862843596022, + 109.38663897969919, + 114.87514690423545, + 110.71243384983525, + 97.51259044999823, + 72.39778902095841, + 24.489905801095226, + 49.07917271377307, + 41.086084102999116, + 38.42291404644562, + 51.15325392240629, + 65.85759198446279, + 76.76583916930579, + 81.6257913697671, + 80.20353801031538, + 72.20311495082414, + 61.19170276598627, + 63.28109879404786 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 851.3189066921525, + 1326.5518535129859, + 1609.221044290723, + 1742.5809808365116, + 1746.6853738523198, + 1632.1026107358648, + 1413.2811207319949, + 1161.4009118822155, + 1215.9340325272856, + 1561.6328726709219, + 1923.9122801839508, + 2226.661242514924, + 2448.080017465779, + 2577.336010519308, + 2604.558954800209, + 2520.376100812142, + 2310.248440388663, + 1951.3192041531165, + 1389.5216274046845 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 4.91488994454613, + 9.948503450916574, + 15.11926481205003, + 20.241335802672026, + 25.124830985471814, + 30.035137787030546, + 35.078322511816275, + 40.07014533759876, + 44.9426815444261, + 49.6931115714545, + 54.12026828862763, + 58.35462095581471, + 62.81674257469332, + 67.51121552820844, + 72.4076611411818, + 77.69495973538872, + 82.53481296075533, + 88.40831545916754, + 91.24919098584027 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 145.36548795887333, + 280.411730240581, + 410.2865542050552, + 536.09591403514, + 658.0460032028413, + 775.9725259275932, + 889.6302925659251, + 998.6251910411793, + 1102.7752220692698, + 1201.1814439113527, + 1292.647877761528, + 1375.3670481007794, + 1447.1915914696597, + 1505.65883141304, + 1544.2728703297594, + 1556.0701090240518, + 1526.66017664746, + 1428.4784357764186, + 1186.3300439504387 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 242.2009251811794, + 226.78423066795636, + 211.42818424761032, + 196.14703057240666, + 180.95974360075425, + 165.89210559660626, + 150.97994333038858, + 136.2743063470363, + 121.84999198289339, + 107.81996668355704, + 94.36027012785775, + 81.75308333700983, + 70.45752894588836, + 61.204173295291646, + 55.03298071071827, + 53.031102162780925, + 55.650312854818324, + 62.3105732490546, + 71.89748569984397 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 1966.6652406524117, + 1968.6733507276683, + 1970.6814608029247, + 1972.6895708781813, + 1974.6976809534376, + 1976.7057910286935, + 1978.7139011039508, + 1980.7220111792071, + 1982.7301212544637, + 1984.7382313297198, + 1986.7463414049762, + 1988.7544514802328, + 1990.7625615554891, + 1992.7706716307457, + 1994.778781706002, + 1996.7868917812586, + 1998.7950018565148, + 2000.8031119317711, + 2002.8112220070282 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 266.34117800466385, + 253.38824166473228, + 231.15110026372184, + 205.8384739082198, + 179.88970809464792, + 154.26718277770672, + 129.39218067802256, + 105.50781777859422, + 83.89503130821453, + 65.7117562885142, + 53.414829272215954, + 49.69904604985873, + 54.860506687595695, + 66.83410550511452, + 80.48103879439213, + 94.99272759032986, + 109.22312522440023, + 122.50664828922461, + 135.32315717220257 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 371.5259677070179, + 532.4889352860997, + 670.1948056871282, + 793.8261486813573, + 906.0296681680782, + 1008.7435490980872, + 1103.321773520055, + 1190.6293059607194, + 1270.9509298679063, + 1344.7670541012708, + 1411.6253748651386, + 1471.2630880836289, + 1523.7629486014057, + 1569.196893669104, + 1606.055629435565, + 1633.0053395677787, + 1648.4900320672273, + 1648.2520695118499, + 1622.041442428474 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1357.210239966107, + "y1": 1357.210239966107 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1800.6696624197764, + "y1": 1800.6696624197764 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1045.3193341911024, + "y1": 1045.3193341911024 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1984.7382313297196, + "y1": 1984.7382313297196 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 1227.6932082267313, + "y1": 1227.6932082267313 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "News Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "\n", + "news_donor, news_receiver = train_test_split(\n", + " df_news, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "news_receiver_no_target = news_receiver.drop(columns=[news_target])\n", + "\n", + "print(f\"Donor size: {len(news_donor)}\")\n", + "print(f\"Receiver size: {len(news_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute with top predictors...\")\n", + "news_result = autoimpute(\n", + " donor_data=news_donor,\n", + " receiver_data=news_receiver_no_target.copy(),\n", + " predictors=news_predictors,\n", + " imputed_variables=[news_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=news_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"News Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 32, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (Online News) ---\n", + "QuantReg: Wasserstein Distance = 2376.8148\n", + "QRF: Wasserstein Distance = 783.7171\n", + "OLS: Wasserstein Distance = 2777.1152\n", + "Matching: Wasserstein Distance = 239.8596\n", + "MDN: Wasserstein Distance = 2038.0328\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
VariableMetricDistance
0shareswasserstein_distance2376.814836
\n", + "
" + ], + "text/plain": [ + " Variable Metric Distance\n", + "0 shares wasserstein_distance 2376.814836" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (Online News) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "news_wasserstein = {}\n", + "\n", + "distribution_comparison_news = compare_distributions(\n", + " donor_data=news_receiver, # Ground truth test split\n", + " receiver_data=news_result.receiver_data, # Imputed values\n", + " imputed_variables=[news_target],\n", + ")\n", + "best_method_name = news_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "news_wasserstein[best_method_name] = distribution_comparison_news[\n", + " distribution_comparison_news['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {news_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in news_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = news_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [news_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=news_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[news_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " news_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {news_wasserstein.get(method_name, 'N/A'):.4f}\")\n", + " \n", + "# Display full comparison for best method \n", + "display(distribution_comparison_news)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 48, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "53b3b2254999454281449d68d51816db", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/9 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
SexLengthDiameterHeightWhole_weightShucked_weightViscera_weightShell_weightClass_number_of_rings
0M0.4550.3650.0950.51400.22450.10100.15015
1M0.3500.2650.0900.22550.09950.04850.0707
2F0.5300.4200.1350.67700.25650.14150.2109
3M0.4400.3650.1250.51600.21550.11400.15510
4I0.3300.2550.0800.20500.08950.03950.0557
\n", + "" + ], + "text/plain": [ + " Sex Length Diameter Height Whole_weight Shucked_weight Viscera_weight \\\n", + "0 M 0.455 0.365 0.095 0.5140 0.2245 0.1010 \n", + "1 M 0.350 0.265 0.090 0.2255 0.0995 0.0485 \n", + "2 F 0.530 0.420 0.135 0.6770 0.2565 0.1415 \n", + "3 M 0.440 0.365 0.125 0.5160 0.2155 0.1140 \n", + "4 I 0.330 0.255 0.080 0.2050 0.0895 0.0395 \n", + "\n", + " Shell_weight Class_number_of_rings \n", + "0 0.150 15 \n", + "1 0.070 7 \n", + "2 0.210 9 \n", + "3 0.155 10 \n", + "4 0.055 7 " + ] + }, + "execution_count": 33, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_abalone = load_dataset(\"abalone\")\n", + "print(f\"\\nShape: {df_abalone.shape}\")\n", + "print(f\"\\nColumn names:\")\n", + "for i, col in enumerate(df_abalone.columns):\n", + " print(f\" {i+1}. {col}\")\n", + "print(f\"\\nData types:\")\n", + "print(df_abalone.dtypes)\n", + "print(f\"\\nFirst few rows:\")\n", + "df_abalone.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 34, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: Shell_weight\n", + "Predictors (8): ['Sex', 'Length', 'Diameter', 'Height', 'Whole_weight', 'Shucked_weight', 'Viscera_weight', 'Class_number_of_rings']\n" + ] + } + ], + "source": [ + "abalone_target = \"Shell_weight\" # Target variable to impute\n", + "abalone_predictors = [col for col in df_abalone.columns if col != abalone_target]\n", + "\n", + "print(f\"Target variable: {abalone_target}\")\n", + "print(f\"Predictors ({len(abalone_predictors)}): {abalone_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 35, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
SexLengthDiameterHeightWhole_weightShucked_weightViscera_weightClass_number_of_rings
Sex1.000000-0.036066-0.038874-0.042077-0.021391-0.001373-0.032067-0.034627
Length-0.0360661.0000000.9868120.8275540.9252610.8979140.9030180.556720
Diameter-0.0388740.9868121.0000000.8336840.9254520.8931620.8997240.574660
Height-0.0420770.8275540.8336841.0000000.8192210.7749720.7983190.557467
Whole_weight-0.0213910.9252610.9254520.8192211.0000000.9694050.9663750.540390
Shucked_weight-0.0013730.8979140.8931620.7749720.9694051.0000000.9319610.420884
Viscera_weight-0.0320670.9030180.8997240.7983190.9663750.9319611.0000000.503819
Class_number_of_rings-0.0346270.5567200.5746600.5574670.5403900.4208840.5038191.000000
\n", + "
" + ], + "text/plain": [ + " Sex Length Diameter Height Whole_weight \\\n", + "Sex 1.000000 -0.036066 -0.038874 -0.042077 -0.021391 \n", + "Length -0.036066 1.000000 0.986812 0.827554 0.925261 \n", + "Diameter -0.038874 0.986812 1.000000 0.833684 0.925452 \n", + "Height -0.042077 0.827554 0.833684 1.000000 0.819221 \n", + "Whole_weight -0.021391 0.925261 0.925452 0.819221 1.000000 \n", + "Shucked_weight -0.001373 0.897914 0.893162 0.774972 0.969405 \n", + "Viscera_weight -0.032067 0.903018 0.899724 0.798319 0.966375 \n", + "Class_number_of_rings -0.034627 0.556720 0.574660 0.557467 0.540390 \n", + "\n", + " Shucked_weight Viscera_weight Class_number_of_rings \n", + "Sex -0.001373 -0.032067 -0.034627 \n", + "Length 0.897914 0.903018 0.556720 \n", + "Diameter 0.893162 0.899724 0.574660 \n", + "Height 0.774972 0.798319 0.557467 \n", + "Whole_weight 0.969405 0.966375 0.540390 \n", + "Shucked_weight 1.000000 0.931961 0.420884 \n", + "Viscera_weight 0.931961 1.000000 0.503819 \n", + "Class_number_of_rings 0.420884 0.503819 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
SexLengthDiameterHeightWhole_weightShucked_weightViscera_weightClass_number_of_rings
Sex1.000000-0.021516-0.022653-0.029176-0.013197-0.000023-0.021652-0.020349
Length-0.0215161.0000000.9833190.8882060.9726330.9568300.9526580.604385
Diameter-0.0226530.9833191.0000000.8957050.9713240.9504720.9483910.622895
Height-0.0291760.8882060.8957051.0000000.9159850.8741960.9005870.657716
Whole_weight-0.0131970.9726330.9713240.9159851.0000000.9770600.9752520.630832
Shucked_weight-0.0000230.9568300.9504720.8741960.9770601.0000000.9476350.539420
Viscera_weight-0.0216520.9526580.9483910.9005870.9752520.9476351.0000000.614344
Class_number_of_rings-0.0203490.6043850.6228950.6577160.6308320.5394200.6143441.000000
\n", + "
" + ], + "text/plain": [ + " Sex Length Diameter Height Whole_weight \\\n", + "Sex 1.000000 -0.021516 -0.022653 -0.029176 -0.013197 \n", + "Length -0.021516 1.000000 0.983319 0.888206 0.972633 \n", + "Diameter -0.022653 0.983319 1.000000 0.895705 0.971324 \n", + "Height -0.029176 0.888206 0.895705 1.000000 0.915985 \n", + "Whole_weight -0.013197 0.972633 0.971324 0.915985 1.000000 \n", + "Shucked_weight -0.000023 0.956830 0.950472 0.874196 0.977060 \n", + "Viscera_weight -0.021652 0.952658 0.948391 0.900587 0.975252 \n", + "Class_number_of_rings -0.020349 0.604385 0.622895 0.657716 0.630832 \n", + "\n", + " Shucked_weight Viscera_weight Class_number_of_rings \n", + "Sex -0.000023 -0.021652 -0.020349 \n", + "Length 0.956830 0.952658 0.604385 \n", + "Diameter 0.950472 0.948391 0.622895 \n", + "Height 0.874196 0.900587 0.657716 \n", + "Whole_weight 0.977060 0.975252 0.630832 \n", + "Shucked_weight 1.000000 0.947635 0.539420 \n", + "Viscera_weight 0.947635 1.000000 0.614344 \n", + "Class_number_of_rings 0.539420 0.614344 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
SexLengthDiameterHeightWhole_weightShucked_weightViscera_weightClass_number_of_rings
Sex1.0000000.1280150.1187490.1258840.1276970.1257220.1302540.093153
Length0.1280151.0000000.3009490.1831780.2560540.2178210.2045590.105884
Diameter0.1187490.3009491.0000000.1883960.2711220.2163340.2063020.112585
Height0.1258840.1831780.1883961.0000000.2070100.1712740.1909950.101486
Whole_weight0.1276970.2560540.2711220.2070101.0000000.1737320.1741600.108976
Shucked_weight0.1257220.2178210.2163340.1712740.1737321.0000000.1341410.091902
Viscera_weight0.1302540.2045590.2063020.1909950.1741600.1341411.0000000.105824
Class_number_of_rings0.0931530.1058840.1125850.1014860.1089760.0919020.1058241.000000
\n", + "
" + ], + "text/plain": [ + " Sex Length Diameter Height Whole_weight \\\n", + "Sex 1.000000 0.128015 0.118749 0.125884 0.127697 \n", + "Length 0.128015 1.000000 0.300949 0.183178 0.256054 \n", + "Diameter 0.118749 0.300949 1.000000 0.188396 0.271122 \n", + "Height 0.125884 0.183178 0.188396 1.000000 0.207010 \n", + "Whole_weight 0.127697 0.256054 0.271122 0.207010 1.000000 \n", + "Shucked_weight 0.125722 0.217821 0.216334 0.171274 0.173732 \n", + "Viscera_weight 0.130254 0.204559 0.206302 0.190995 0.174160 \n", + "Class_number_of_rings 0.093153 0.105884 0.112585 0.101486 0.108976 \n", + "\n", + " Shucked_weight Viscera_weight Class_number_of_rings \n", + "Sex 0.125722 0.130254 0.093153 \n", + "Length 0.217821 0.204559 0.105884 \n", + "Diameter 0.216334 0.206302 0.112585 \n", + "Height 0.171274 0.190995 0.101486 \n", + "Whole_weight 0.173732 0.174160 0.108976 \n", + "Shucked_weight 1.000000 0.134141 0.091902 \n", + "Viscera_weight 0.134141 1.000000 0.105824 \n", + "Class_number_of_rings 0.091902 0.105824 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
Shell_weight
Whole_weight0.186427
Diameter0.167913
Length0.159534
Viscera_weight0.141901
Shucked_weight0.127999
Height0.118932
Class_number_of_rings0.052894
Sex0.024884
\n", + "
" + ], + "text/plain": [ + " Shell_weight\n", + "Whole_weight 0.186427\n", + "Diameter 0.167913\n", + "Length 0.159534\n", + "Viscera_weight 0.141901\n", + "Shucked_weight 0.127999\n", + "Height 0.118932\n", + "Class_number_of_rings 0.052894\n", + "Sex 0.024884" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "abalone_correlations = compute_predictor_correlations(\n", + " data=df_abalone,\n", + " predictors=abalone_predictors,\n", + " imputed_variables=[abalone_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(abalone_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(abalone_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(abalone_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(abalone_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=abalone_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 36, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "d1f96f060e724322a7cea8f37c286482", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/8 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
4Whole_weight0.01144100.00267930.5729620.0087620
5Shucked_weight0.01008500.00132215.0914410.0087620
6Viscera_weight0.00900000.0002372.7079030.0087620
3Height0.00888000.0001181.3425250.0087620
2Diameter0.00877100.0000090.1024890.0087620
0Sex0.0087550-0.000007-0.0829350.0087620
1Length0.0087140-0.000048-0.5486170.0087620
7Class_number_of_rings0.0086500-0.000112-1.2815980.0087620
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "4 Whole_weight 0.011441 0 0.002679 \n", + "5 Shucked_weight 0.010085 0 0.001322 \n", + "6 Viscera_weight 0.009000 0 0.000237 \n", + "3 Height 0.008880 0 0.000118 \n", + "2 Diameter 0.008771 0 0.000009 \n", + "0 Sex 0.008755 0 -0.000007 \n", + "1 Length 0.008714 0 -0.000048 \n", + "7 Class_number_of_rings 0.008650 0 -0.000112 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "4 30.572962 0.008762 0 \n", + "5 15.091441 0.008762 0 \n", + "6 2.707903 0.008762 0 \n", + "3 1.342525 0.008762 0 \n", + "2 0.102489 0.008762 0 \n", + "0 -0.082935 0.008762 0 \n", + "1 -0.548617 0.008762 0 \n", + "7 -1.281598 0.008762 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis...\")\n", + "abalone_loo = leave_one_out_analysis(\n", + " data=df_abalone,\n", + " predictors=abalone_predictors,\n", + " imputed_variables=[abalone_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(abalone_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 37, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 2506\n", + "Receiver size: 1671\n", + "\n", + "Running autoimpute...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "077feb4af4a6419da4569e772accd507", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.003171187505467154, + 0.005130387233501125, + 0.0070908502119267445, + 0.008298522317913974, + 0.00977020928262996, + 0.010720453793608002, + 0.011885963511224562, + 0.012002792423121883, + 0.012249910279043508, + 0.012238157748248522, + 0.012173357885821981, + 0.011792550357452424, + 0.011332565884167919, + 0.010413367488131308, + 0.009396204403941122, + 0.008270376378716669, + 0.007089934215234869, + 0.005423739532886418, + 0.0036155805719238826 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.000370609330897358, + 0.0003652745273775171, + 0.0003542015967681174, + 0.0003581402999814908, + 0.0003486765666195462, + 0.0003441475733918232, + 0.0003450233024532522, + 0.00036724649880935217, + 0.00039053815097184126, + 0.00041447525982789247, + 0.00045442297546678445, + 0.0004970591309255817, + 0.0005384563278486812, + 0.0005448871576265109, + 0.0005371124042098624, + 0.0005085526161979347, + 0.0004891847358885552, + 0.00047062940171582203, + 0.00045756947185525556 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.003640431971412939, + 0.0054201672745586415, + 0.006631847486593419, + 0.00746375357970813, + 0.007996822864203368, + 0.008308100389038897, + 0.008445008917186806, + 0.008491620640887173, + 0.00851848761584949, + 0.008581834515710474, + 0.008684830402663123, + 0.008760733491948555, + 0.00873043495812654, + 0.008578226744835217, + 0.008254303974197876, + 0.0076962383272222035, + 0.006860191812274187, + 0.005653547347567274, + 0.0038813771491098977 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.0003002957405243558, + 0.0003916647076410419, + 0.0004575980596868312, + 0.00048010456147152734, + 0.0004743316771156946, + 0.00046776978159706847, + 0.00046703103575669147, + 0.0004901431426791614, + 0.0005164720063186571, + 0.000565445067077025, + 0.0005982200297485073, + 0.0006221877855910843, + 0.0006330517928663067, + 0.00064919727119191, + 0.0006620420285619062, + 0.0006772332349061551, + 0.0006710739189733639, + 0.0006771438751547985, + 0.0006343251427314907 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.0025812372039487864, + 0.00415203322551073, + 0.00536652434879917, + 0.006295003165407054, + 0.0069760216580269764, + 0.007467090953206039, + 0.007856627683712415, + 0.008124981770825527, + 0.008272805362843293, + 0.008280681178693978, + 0.008220785772956412, + 0.008096860802368306, + 0.007803099764698377, + 0.00739903476423837, + 0.0068958004097455575, + 0.0062701701281025505, + 0.005462830005979865, + 0.0044332588930873825, + 0.003138288530458782 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.0020365104336603802, + 0.0019380406463349053, + 0.0018407949105962973, + 0.0017449778836709257, + 0.0016508383572344301, + 0.0015586803086344201, + 0.0014688767419103995, + 0.0013818867598028154, + 0.0012982760517331199, + 0.0012187403134718854, + 0.0011441296921467239, + 0.0010754697154333852, + 0.001013969930900011, + 0.0009610059772534084, + 0.0009180563770578484, + 0.0008865777406346185, + 0.0008678192349293623, + 0.0008626111026588879, + 0.0008711963957339234 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.014285100178129794, + 0.014380612201891041, + 0.014476124225652282, + 0.014571636249413525, + 0.014667148273174765, + 0.01476266029693601, + 0.014858172320697253, + 0.014953684344458493, + 0.015049196368219737, + 0.015144708391980977, + 0.015240220415742222, + 0.015335732439503463, + 0.015431244463264706, + 0.015526756487025947, + 0.015622268510787189, + 0.015717780534548434, + 0.015813292558309676, + 0.015908804582070918, + 0.01600431660583216 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 0.004933235896485864, + 0.0061496603531529725, + 0.007025759919520332, + 0.007787472230557316, + 0.008366206347179553, + 0.008728138651863643, + 0.00893846297741526, + 0.008968634550059822, + 0.008847267828909986, + 0.008667859487156317, + 0.008372227090893001, + 0.007942799639520893, + 0.007393834191791668, + 0.006813615496239677, + 0.006038886484706126, + 0.005016465552396074, + 0.003888893897177334, + 0.002629321426474522, + 0.0016051762364270791 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 0.012276574516477503, + 0.018432648819518734, + 0.02317406481400918, + 0.02672731973268183, + 0.029399918741729713, + 0.03134876719068555, + 0.03275305750018242, + 0.03354567398997228, + 0.03382556654362355, + 0.033535913691053534, + 0.03283750432650534, + 0.031678071755279626, + 0.030068237694899587, + 0.028108862716910155, + 0.02571725838163568, + 0.02272632121067506, + 0.01919042724359827, + 0.014768039973995072, + 0.009304198960612093 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.009056111106576949, + "y1": 0.009056111106576949 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.0073998926033207475, + "y1": 0.0073998926033207475 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.006478586085400504, + "y1": 0.006478586085400504 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.015144708391980977, + "y1": 0.015144708391980977 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 0.02575886462126553, + "y1": 0.02575886462126553 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Abalone Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "abalone_donor, abalone_receiver = train_test_split(\n", + " df_abalone, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "abalone_receiver_no_target = abalone_receiver.drop(columns=[abalone_target])\n", + "\n", + "print(f\"Donor size: {len(abalone_donor)}\")\n", + "print(f\"Receiver size: {len(abalone_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute...\")\n", + "abalone_result = autoimpute(\n", + " donor_data=abalone_donor,\n", + " receiver_data=abalone_receiver_no_target.copy(),\n", + " predictors=abalone_predictors,\n", + " imputed_variables=[abalone_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=abalone_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"Abalone Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 38, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (Abalone) ---\n", + "QuantReg: Wasserstein Distance = 0.0027\n", + "QRF: Wasserstein Distance = 0.0028\n", + "OLS: Wasserstein Distance = 0.0037\n", + "Matching: Wasserstein Distance = 0.0025\n", + "MDN: Wasserstein Distance = 0.0404\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
VariableMetricDistance
0Shell_weightwasserstein_distance0.002713
\n", + "
" + ], + "text/plain": [ + " Variable Metric Distance\n", + "0 Shell_weight wasserstein_distance 0.002713" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (Abalone) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "abalone_wasserstein = {}\n", + "\n", + "distribution_comparison_abalone = compare_distributions(\n", + " donor_data=abalone_receiver, # Ground truth test split\n", + " receiver_data=abalone_result.receiver_data, # Imputed values\n", + " imputed_variables=[abalone_target],\n", + ")\n", + "best_method_name = abalone_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "abalone_wasserstein[best_method_name] = distribution_comparison_abalone[\n", + " distribution_comparison_abalone['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {abalone_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in abalone_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = abalone_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [abalone_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=abalone_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[abalone_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " abalone_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {abalone_wasserstein.get(method_name, 'N/A'):.4f}\")\n", + " \n", + "# Display full comparison for best method\n", + "display(distribution_comparison_abalone)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 39, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "b4b2625d9d3f48199e2a8917540755cb", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/8 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
bedroomsbathroomssqft_livingsqft_lotfloorswaterfrontviewconditiongradesqft_abovesqft_basementyr_builtyr_renovatedzipcodelatlongsqft_living15sqft_lot15date_yeardate_monthdate_dayprice
031.00118056501.0003711800195509817847.5112-122.2571340565020141013221900.0
132.25257072422.000372170400195119919812547.7210-122.319169076392014129538000.0
221.00770100001.000367700193309802847.7379-122.233272080622015225180000.0
343.00196050001.000571050910196509813647.5208-122.393136050002014129604000.0
432.00168080801.0003816800198709807447.6168-122.045180075032015218510000.0
\n", + "" + ], + "text/plain": [ + " bedrooms bathrooms sqft_living sqft_lot floors waterfront view \\\n", + "0 3 1.00 1180 5650 1.0 0 0 \n", + "1 3 2.25 2570 7242 2.0 0 0 \n", + "2 2 1.00 770 10000 1.0 0 0 \n", + "3 4 3.00 1960 5000 1.0 0 0 \n", + "4 3 2.00 1680 8080 1.0 0 0 \n", + "\n", + " condition grade sqft_above sqft_basement yr_built yr_renovated \\\n", + "0 3 7 1180 0 1955 0 \n", + "1 3 7 2170 400 1951 1991 \n", + "2 3 6 770 0 1933 0 \n", + "3 5 7 1050 910 1965 0 \n", + "4 3 8 1680 0 1987 0 \n", + "\n", + " zipcode lat long sqft_living15 sqft_lot15 date_year \\\n", + "0 98178 47.5112 -122.257 1340 5650 2014 \n", + "1 98125 47.7210 -122.319 1690 7639 2014 \n", + "2 98028 47.7379 -122.233 2720 8062 2015 \n", + "3 98136 47.5208 -122.393 1360 5000 2014 \n", + "4 98074 47.6168 -122.045 1800 7503 2015 \n", + "\n", + " date_month date_day price \n", + "0 10 13 221900.0 \n", + "1 12 9 538000.0 \n", + "2 2 25 180000.0 \n", + "3 12 9 604000.0 \n", + "4 2 18 510000.0 " + ] + }, + "execution_count": 40, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "df_house = load_dataset(\"house_sales\")\n", + "print(f\"\\nShape: {df_house.shape}\")\n", + "print(f\"\\nColumn names:\")\n", + "for i, col in enumerate(df_house.columns):\n", + " print(f\" {i+1}. {col}\")\n", + "print(f\"\\nData types:\")\n", + "print(df_house.dtypes)\n", + "print(f\"\\nFirst few rows:\")\n", + "df_house.head()" + ] + }, + { + "cell_type": "code", + "execution_count": 41, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Target variable: price\n", + "Predictors (10): ['bedrooms', 'bathrooms', 'sqft_living', 'sqft_lot', 'waterfront', 'view', 'condition', 'grade', 'sqft_above', 'sqft_basement']\n" + ] + } + ], + "source": [ + "house_target = \"price\" # Target variable to impute\n", + "house_predictors = [\"bedrooms\", \"bathrooms\", \"sqft_living\", \"sqft_lot\", \"waterfront\", \"view\", \"condition\", \"grade\", \"sqft_above\", \"sqft_basement\"]\n", + "\n", + "print(f\"Target variable: {house_target}\")\n", + "print(f\"Predictors ({len(house_predictors)}): {house_predictors}\")" + ] + }, + { + "cell_type": "code", + "execution_count": 42, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Computing predictor correlations...\n", + "\n", + "--- Pearson Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
bedroomsbathroomssqft_livingsqft_lotwaterfrontviewconditiongradesqft_abovesqft_basement
bedrooms1.0000000.5158840.5766710.031703-0.0065820.0795320.0284720.3569670.4776000.303093
bathrooms0.5158841.0000000.7546650.0877400.0637440.187737-0.1249820.6649830.6853420.283770
sqft_living0.5766710.7546651.0000000.1728260.1038180.284611-0.0587530.7627040.8765970.435043
sqft_lot0.0317030.0877400.1728261.0000000.0216040.074710-0.0089580.1136210.1835120.015286
waterfront-0.0065820.0637440.1038180.0216041.0000000.4018570.0166530.0827750.0720750.080588
view0.0795320.1877370.2846110.0747100.4018571.0000000.0459900.2513210.1676490.276947
condition0.028472-0.124982-0.058753-0.0089580.0166530.0459901.000000-0.144674-0.1582140.174105
grade0.3569670.6649830.7627040.1136210.0827750.251321-0.1446741.0000000.7559230.168392
sqft_above0.4776000.6853420.8765970.1835120.0720750.167649-0.1582140.7559231.000000-0.051943
sqft_basement0.3030930.2837700.4350430.0152860.0805880.2769470.1741050.168392-0.0519431.000000
\n", + "
" + ], + "text/plain": [ + " bedrooms bathrooms sqft_living sqft_lot waterfront \\\n", + "bedrooms 1.000000 0.515884 0.576671 0.031703 -0.006582 \n", + "bathrooms 0.515884 1.000000 0.754665 0.087740 0.063744 \n", + "sqft_living 0.576671 0.754665 1.000000 0.172826 0.103818 \n", + "sqft_lot 0.031703 0.087740 0.172826 1.000000 0.021604 \n", + "waterfront -0.006582 0.063744 0.103818 0.021604 1.000000 \n", + "view 0.079532 0.187737 0.284611 0.074710 0.401857 \n", + "condition 0.028472 -0.124982 -0.058753 -0.008958 0.016653 \n", + "grade 0.356967 0.664983 0.762704 0.113621 0.082775 \n", + "sqft_above 0.477600 0.685342 0.876597 0.183512 0.072075 \n", + "sqft_basement 0.303093 0.283770 0.435043 0.015286 0.080588 \n", + "\n", + " view condition grade sqft_above sqft_basement \n", + "bedrooms 0.079532 0.028472 0.356967 0.477600 0.303093 \n", + "bathrooms 0.187737 -0.124982 0.664983 0.685342 0.283770 \n", + "sqft_living 0.284611 -0.058753 0.762704 0.876597 0.435043 \n", + "sqft_lot 0.074710 -0.008958 0.113621 0.183512 0.015286 \n", + "waterfront 0.401857 0.016653 0.082775 0.072075 0.080588 \n", + "view 1.000000 0.045990 0.251321 0.167649 0.276947 \n", + "condition 0.045990 1.000000 -0.144674 -0.158214 0.174105 \n", + "grade 0.251321 -0.144674 1.000000 0.755923 0.168392 \n", + "sqft_above 0.167649 -0.158214 0.755923 1.000000 -0.051943 \n", + "sqft_basement 0.276947 0.174105 0.168392 -0.051943 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Spearman Correlation Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
bedroomsbathroomssqft_livingsqft_lotwaterfrontviewconditiongradesqft_abovesqft_basement
bedrooms1.0000000.5214060.6473660.216531-0.0075980.0814210.0126600.3805340.5397040.230496
bathrooms0.5214061.0000000.7455260.0688050.0495220.155549-0.1628910.6581940.6910060.191848
sqft_living0.6473660.7455261.0000000.3041590.0703260.232994-0.0626380.7164000.8435040.327878
sqft_lot0.2165310.0688050.3041591.0000000.0856010.1170330.1147240.1520490.2724080.036624
waterfront-0.0075980.0495220.0703260.0856011.0000000.2849240.0167440.0621890.0544900.051969
view0.0814210.1555490.2329940.1170330.2849241.0000000.0460000.2170440.1443940.236525
condition0.012660-0.162891-0.0626380.1147240.0167440.0460001.000000-0.167374-0.1581260.161623
grade0.3805340.6581940.7164000.1520490.0621890.217044-0.1673741.0000000.7118150.092927
sqft_above0.5397040.6910060.8435040.2724080.0544900.144394-0.1581260.7118151.000000-0.165644
sqft_basement0.2304960.1918480.3278780.0366240.0519690.2365250.1616230.092927-0.1656441.000000
\n", + "
" + ], + "text/plain": [ + " bedrooms bathrooms sqft_living sqft_lot waterfront \\\n", + "bedrooms 1.000000 0.521406 0.647366 0.216531 -0.007598 \n", + "bathrooms 0.521406 1.000000 0.745526 0.068805 0.049522 \n", + "sqft_living 0.647366 0.745526 1.000000 0.304159 0.070326 \n", + "sqft_lot 0.216531 0.068805 0.304159 1.000000 0.085601 \n", + "waterfront -0.007598 0.049522 0.070326 0.085601 1.000000 \n", + "view 0.081421 0.155549 0.232994 0.117033 0.284924 \n", + "condition 0.012660 -0.162891 -0.062638 0.114724 0.016744 \n", + "grade 0.380534 0.658194 0.716400 0.152049 0.062189 \n", + "sqft_above 0.539704 0.691006 0.843504 0.272408 0.054490 \n", + "sqft_basement 0.230496 0.191848 0.327878 0.036624 0.051969 \n", + "\n", + " view condition grade sqft_above sqft_basement \n", + "bedrooms 0.081421 0.012660 0.380534 0.539704 0.230496 \n", + "bathrooms 0.155549 -0.162891 0.658194 0.691006 0.191848 \n", + "sqft_living 0.232994 -0.062638 0.716400 0.843504 0.327878 \n", + "sqft_lot 0.117033 0.114724 0.152049 0.272408 0.036624 \n", + "waterfront 0.284924 0.016744 0.062189 0.054490 0.051969 \n", + "view 1.000000 0.046000 0.217044 0.144394 0.236525 \n", + "condition 0.046000 1.000000 -0.167374 -0.158126 0.161623 \n", + "grade 0.217044 -0.167374 1.000000 0.711815 0.092927 \n", + "sqft_above 0.144394 -0.158126 0.711815 1.000000 -0.165644 \n", + "sqft_basement 0.236525 0.161623 0.092927 -0.165644 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Mutual Information Matrix ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
bedroomsbathroomssqft_livingsqft_lotwaterfrontviewconditiongradesqft_abovesqft_basement
bedrooms1.0000000.0981990.1591970.0383200.0373470.0076490.0063230.0493920.1124570.034205
bathrooms0.0981991.0000000.1511310.0286460.0560600.0298640.0340500.1488060.1176330.030131
sqft_living0.1591970.1511311.0000000.0152660.0496430.0471290.0107420.2004840.4051670.087963
sqft_lot0.0383200.0286460.0152661.0000000.0812740.0313700.0338650.0430080.0168370.019837
waterfront0.0373470.0560600.0496430.0812741.0000000.4180370.0000000.0507050.0287530.215099
view0.0076490.0298640.0471290.0313700.4180371.0000000.0000000.0434770.0239200.074452
condition0.0063230.0340500.0107420.0338650.0000000.0000001.0000000.0206810.0203480.018175
grade0.0493920.1488060.2004840.0430080.0507050.0434770.0206811.0000000.1931970.020057
sqft_above0.1124570.1176330.4051670.0168370.0287530.0239200.0203480.1931971.0000000.043748
sqft_basement0.0342050.0301310.0879630.0198370.2150990.0744520.0181750.0200570.0437481.000000
\n", + "
" + ], + "text/plain": [ + " bedrooms bathrooms sqft_living sqft_lot waterfront \\\n", + "bedrooms 1.000000 0.098199 0.159197 0.038320 0.037347 \n", + "bathrooms 0.098199 1.000000 0.151131 0.028646 0.056060 \n", + "sqft_living 0.159197 0.151131 1.000000 0.015266 0.049643 \n", + "sqft_lot 0.038320 0.028646 0.015266 1.000000 0.081274 \n", + "waterfront 0.037347 0.056060 0.049643 0.081274 1.000000 \n", + "view 0.007649 0.029864 0.047129 0.031370 0.418037 \n", + "condition 0.006323 0.034050 0.010742 0.033865 0.000000 \n", + "grade 0.049392 0.148806 0.200484 0.043008 0.050705 \n", + "sqft_above 0.112457 0.117633 0.405167 0.016837 0.028753 \n", + "sqft_basement 0.034205 0.030131 0.087963 0.019837 0.215099 \n", + "\n", + " view condition grade sqft_above sqft_basement \n", + "bedrooms 0.007649 0.006323 0.049392 0.112457 0.034205 \n", + "bathrooms 0.029864 0.034050 0.148806 0.117633 0.030131 \n", + "sqft_living 0.047129 0.010742 0.200484 0.405167 0.087963 \n", + "sqft_lot 0.031370 0.033865 0.043008 0.016837 0.019837 \n", + "waterfront 0.418037 0.000000 0.050705 0.028753 0.215099 \n", + "view 1.000000 0.000000 0.043477 0.023920 0.074452 \n", + "condition 0.000000 1.000000 0.020681 0.020348 0.018175 \n", + "grade 0.043477 0.020681 1.000000 0.193197 0.020057 \n", + "sqft_above 0.023920 0.020348 0.193197 1.000000 0.043748 \n", + "sqft_basement 0.074452 0.018175 0.020057 0.043748 1.000000 " + ] + }, + "metadata": {}, + "output_type": "display_data" + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Predictor-Target Mutual Information ---\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
price
sqft_living0.034919
grade0.033545
sqft_above0.025859
bathrooms0.020110
bedrooms0.007915
sqft_basement0.006930
sqft_lot0.006063
view0.005654
condition0.001584
waterfront0.001340
\n", + "
" + ], + "text/plain": [ + " price\n", + "sqft_living 0.034919\n", + "grade 0.033545\n", + "sqft_above 0.025859\n", + "bathrooms 0.020110\n", + "bedrooms 0.007915\n", + "sqft_basement 0.006930\n", + "sqft_lot 0.006063\n", + "view 0.005654\n", + "condition 0.001584\n", + "waterfront 0.001340" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Computing predictor correlations...\")\n", + "house_correlations = compute_predictor_correlations(\n", + " data=df_house,\n", + " predictors=house_predictors,\n", + " imputed_variables=[house_target],\n", + " method=\"all\"\n", + ")\n", + "\n", + "print(\"\\n--- Pearson Correlation Matrix ---\")\n", + "display(house_correlations[\"pearson\"])\n", + "\n", + "print(\"\\n--- Spearman Correlation Matrix ---\")\n", + "display(house_correlations[\"spearman\"])\n", + "\n", + "print(\"\\n--- Mutual Information Matrix ---\")\n", + "display(house_correlations[\"mutual_info\"])\n", + "\n", + "print(\"\\n--- Predictor-Target Mutual Information ---\")\n", + "display(house_correlations[\"predictor_target_mi\"].sort_values(\n", + " by=house_target, ascending=False\n", + "))" + ] + }, + { + "cell_type": "code", + "execution_count": 43, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running leave-one-out analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "9424f4e64a4c41d4b1c132f0ae055ee0", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Leave-one-out analysis: 0%| | 0/10 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
predictor_removedavg_quantile_lossavg_log_lossloss_increaserelative_impactbaseline_quantile_lossbaseline_log_loss
7grade69216.84791206095.0555999.65602463121.7923130
3sqft_lot68609.38990805487.5975958.69366663121.7923130
6condition64704.66252401582.8702122.50764563121.7923130
1bathrooms64190.97316001069.1808481.69383863121.7923130
5view63788.4725350666.6802221.05618163121.7923130
0bedrooms63643.8861950522.0938820.82712163121.7923130
2sqft_living63438.1119780316.3196660.50112663121.7923130
4waterfront63387.5467790265.7544670.42101963121.7923130
9sqft_basement63230.8761540109.0838420.17281563121.7923130
8sqft_above63195.338181073.5458680.11651463121.7923130
\n", + "" + ], + "text/plain": [ + " predictor_removed avg_quantile_loss avg_log_loss loss_increase \\\n", + "7 grade 69216.847912 0 6095.055599 \n", + "3 sqft_lot 68609.389908 0 5487.597595 \n", + "6 condition 64704.662524 0 1582.870212 \n", + "1 bathrooms 64190.973160 0 1069.180848 \n", + "5 view 63788.472535 0 666.680222 \n", + "0 bedrooms 63643.886195 0 522.093882 \n", + "2 sqft_living 63438.111978 0 316.319666 \n", + "4 waterfront 63387.546779 0 265.754467 \n", + "9 sqft_basement 63230.876154 0 109.083842 \n", + "8 sqft_above 63195.338181 0 73.545868 \n", + "\n", + " relative_impact baseline_quantile_loss baseline_log_loss \n", + "7 9.656024 63121.792313 0 \n", + "3 8.693666 63121.792313 0 \n", + "6 2.507645 63121.792313 0 \n", + "1 1.693838 63121.792313 0 \n", + "5 1.056181 63121.792313 0 \n", + "0 0.827121 63121.792313 0 \n", + "2 0.501126 63121.792313 0 \n", + "4 0.421019 63121.792313 0 \n", + "9 0.172815 63121.792313 0 \n", + "8 0.116514 63121.792313 0 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Running leave-one-out analysis...\")\n", + "house_loo = leave_one_out_analysis(\n", + " data=df_house,\n", + " predictors=house_predictors,\n", + " imputed_variables=[house_target],\n", + " model_class=QRF,\n", + " train_size=0.6,\n", + " n_jobs=1\n", + ")\n", + "\n", + "print(\"\\n--- Leave-One-Out Results ---\")\n", + "display(house_loo)" + ] + }, + { + "cell_type": "code", + "execution_count": 44, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Creating 60/40 donor/receiver split...\n", + "Donor size: 12967\n", + "Receiver size: 8646\n", + "\n", + "Running autoimpute...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "ff390d112d1f439eb79aa91fc5a0cd3d", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "AutoImputation progress: 0%| | 0/5 [00:00Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 20709.781289760314, + 32711.011025779328, + 44535.18256502814, + 53703.12974447841, + 62417.06348529387, + 68618.35321517424, + 75277.79127833492, + 78690.66612006526, + 81692.4921553432, + 82387.89194252898, + 82190.59544011946, + 80668.23840043221, + 76139.77789636323, + 71953.78781378965, + 66902.27469114999, + 59408.259501290624, + 51162.40596579338, + 40201.64217328191, + 28726.94975485272 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 194.028361567845, + 270.40611376212814, + 424.35766919963396, + 655.5625728964469, + 859.3968486998843, + 1081.5312202798207, + 1290.8863186105052, + 1397.2462643130293, + 1540.027505583281, + 1619.1472241056717, + 1689.5007427358473, + 1708.6528162176992, + 1692.2362737940518, + 1724.0607813481622, + 1740.5827151189064, + 1841.4550330271604, + 1938.1655872462368, + 2037.0636749212763, + 2092.754289131688 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 20473.589998960033, + 33432.25165704795, + 43320.90225158931, + 51196.22021063475, + 57569.70124450637, + 62801.63397552215, + 67045.07770711477, + 70407.69702574884, + 72927.84248013695, + 74521.04034750645, + 75226.70101567755, + 75023.80226665984, + 73828.67718656275, + 71527.86883635657, + 67858.72913281275, + 62484.32419192977, + 54953.553220199516, + 44734.247783678955, + 30447.100712490104 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 182.3720945383094, + 472.3398849856401, + 524.3005985089421, + 971.9646125139536, + 1224.9792826088424, + 2115.933780108482, + 1056.3153214213212, + 1829.0166203736255, + 2127.8709907744037, + 1718.448175433048, + 1352.7976079507, + 1559.8561647881027, + 1275.6384343952695, + 2065.063847283456, + 1573.1974067010156, + 1647.5556756086503, + 1912.659852465349, + 1602.9127058010229, + 1420.088344050314 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 14181.66310174412, + 25755.579151254104, + 35955.151863432664, + 44566.35401926348, + 52056.62402430021, + 58865.9289442663, + 63653.54425385364, + 67731.13211630892, + 70620.50999472741, + 72451.6790063794, + 73581.67597367859, + 72803.27067734861, + 72027.60910654339, + 69655.74924359658, + 65282.60359285939, + 60793.11896628221, + 53597.61821135791, + 42022.121677457835, + 27454.727576089623 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 1980.2054753191617, + 1703.6432646631138, + 1455.5825320961067, + 1253.0658913312486, + 1121.0536267527482, + 1085.579027918468, + 1155.5673513130084, + 1314.2767846796703, + 1534.41992812871, + 1793.515835082672, + 2077.0380524461475, + 2376.259366670552, + 2685.938117275419, + 3002.840663062175, + 3324.9021055377766, + 3650.75738181033, + 3979.474638195215, + 4310.39914822265, + 4643.05898344198 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 101408.0160260291, + 101571.7732345491, + 101735.5304430691, + 101899.2876515891, + 102063.0448601091, + 102226.80206862911, + 102390.5592771491, + 102554.31648566911, + 102718.07369418911, + 102881.83090270912, + 103045.58811122912, + 103209.34531974913, + 103373.10252826913, + 103536.85973678913, + 103700.61694530913, + 103864.37415382915, + 104028.13136234913, + 104191.88857086914, + 104355.64577938915 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 284.7326924435361, + 569.2486180173503, + 853.6520482378114, + 1137.9652075982785, + 1422.2045123242642, + 1706.368936967098, + 1990.47200208671, + 2274.5075204441405, + 2558.4736744376673, + 2842.3662015453397, + 3126.183792588692, + 3409.9218109892086, + 3693.5727944433816, + 3977.115932312908, + 4260.527678263926, + 4543.760410967098, + 4826.740852245481, + 5109.3345422247, + 5391.098581809968 + ] + }, + "hovertemplate": "Method=MDN
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "MDN", + "marker": { + "color": "#332288", + "pattern": { + "shape": "" + } + }, + "name": "MDN", + "offsetgroup": "MDN", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 26871.173484862265, + 53740.50929097005, + 80608.89541027459, + 107476.54579793694, + 134343.55265796493, + 161209.97167135123, + 188075.82470534535, + 214941.1126265573, + 241805.8201917059, + 268669.9238017856, + 295533.39161048527, + 322396.1651101296, + 349258.13342021353, + 376119.19087407645, + 402979.11892909755, + 429837.5833290928, + 456694.0042487571, + 483547.1238196845, + 510393.3069950899 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 60952.489182045254, + "y1": 60952.489182045254 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 58409.52427605975, + "y1": 58409.52427605975 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 54897.71902635497, + "y1": 54897.71902635497 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 102881.83090270912, + "y1": 102881.83090270912 + }, + { + "line": { + "color": "#332288", + "dash": "dot", + "width": 2 + }, + "name": "MDN Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 268657.9656829148, + "y1": 268657.9656829148 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "House Sales Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"Creating 60/40 donor/receiver split...\")\n", + "house_donor, house_receiver = train_test_split(\n", + " df_house, train_size=0.6, random_state=42\n", + ")\n", + "\n", + "house_receiver_no_target = house_receiver.drop(columns=[house_target])\n", + "\n", + "print(f\"Donor size: {len(house_donor)}\")\n", + "print(f\"Receiver size: {len(house_receiver)}\")\n", + "\n", + "print(\"\\nRunning autoimpute...\")\n", + "house_result = autoimpute(\n", + " donor_data=house_donor,\n", + " receiver_data=house_receiver_no_target.copy(),\n", + " predictors=house_predictors,\n", + " imputed_variables=[house_target],\n", + " impute_all=True,\n", + " log_level=\"INFO\"\n", + ")\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=house_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"House Sales Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 45, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Autoimpute CV Results ---\n" + ] + }, + { + "data": { + "application/vnd.plotly.v1+json": { + "config": { + "plotlyServerURL": "https://plot.ly" + }, + "data": [ + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 719.328189753778, + 940.3832731105417, + 1482.848168298151, + 981.7573126166058, + 893.1098804459698, + 890.1490739996166, + 918.5323509728148, + 1588.7521448227917, + 1201.8109300236083, + 1336.585972042465, + 666.4767580846618, + 758.9475683984272, + 1596.3561562877353, + 1413.758832331186, + 1812.523234457851, + 1962.6534657951277, + 1475.01168353744, + 1950.596397145942, + 1745.9008891730946 + ] + }, + "hovertemplate": "Method=QRF
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QRF", + "marker": { + "color": "#88CCEE", + "pattern": { + "shape": "" + } + }, + "name": "QRF", + "offsetgroup": "QRF", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 20709.781289760314, + 32711.011025779328, + 44535.18256502814, + 53703.12974447841, + 62417.06348529387, + 68618.35321517424, + 75277.79127833492, + 78690.66612006526, + 81692.4921553432, + 82387.89194252898, + 82190.59544011946, + 80668.23840043221, + 76139.77789636323, + 71953.78781378965, + 66902.27469114999, + 59408.259501290624, + 51162.40596579338, + 40201.64217328191, + 28726.94975485272 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 194.028361567845, + 270.40611376212814, + 424.35766919963396, + 655.5625728964469, + 859.3968486998843, + 1081.5312202798207, + 1290.8863186105052, + 1397.2462643130293, + 1540.027505583281, + 1619.1472241056717, + 1689.5007427358473, + 1708.6528162176992, + 1692.2362737940518, + 1724.0607813481622, + 1740.5827151189064, + 1841.4550330271604, + 1938.1655872462368, + 2037.0636749212763, + 2092.754289131688 + ] + }, + "hovertemplate": "Method=OLS
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "OLS", + "marker": { + "color": "#CC6677", + "pattern": { + "shape": "" + } + }, + "name": "OLS", + "offsetgroup": "OLS", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 20473.589998960033, + 33432.25165704795, + 43320.90225158931, + 51196.22021063475, + 57569.70124450637, + 62801.63397552215, + 67045.07770711477, + 70407.69702574884, + 72927.84248013695, + 74521.04034750645, + 75226.70101567755, + 75023.80226665984, + 73828.67718656275, + 71527.86883635657, + 67858.72913281275, + 62484.32419192977, + 54953.553220199516, + 44734.247783678955, + 30447.100712490104 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 182.3720945383094, + 472.3398849856401, + 524.3005985089421, + 971.9646125139536, + 1224.9792826088424, + 2115.933780108482, + 1056.3153214213212, + 1829.0166203736255, + 2127.8709907744037, + 1718.448175433048, + 1352.7976079507, + 1559.8561647881027, + 1275.6384343952695, + 2065.063847283456, + 1573.1974067010156, + 1647.5556756086503, + 1912.659852465349, + 1602.9127058010229, + 1420.088344050314 + ] + }, + "hovertemplate": "Method=QuantReg
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "QuantReg", + "marker": { + "color": "#DDCC77", + "pattern": { + "shape": "" + } + }, + "name": "QuantReg", + "offsetgroup": "QuantReg", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 14181.66310174412, + 25755.579151254104, + 35955.151863432664, + 44566.35401926348, + 52056.62402430021, + 58865.9289442663, + 63653.54425385364, + 67731.13211630892, + 70620.50999472741, + 72451.6790063794, + 73581.67597367859, + 72803.27067734861, + 72027.60910654339, + 69655.74924359658, + 65282.60359285939, + 60793.11896628221, + 53597.61821135791, + 42022.121677457835, + 27454.727576089623 + ], + "yaxis": "y" + }, + { + "alignmentgroup": "True", + "error_y": { + "array": [ + 1980.2054753191617, + 1703.6432646631138, + 1455.5825320961067, + 1253.0658913312486, + 1121.0536267527482, + 1085.579027918468, + 1155.5673513130084, + 1314.2767846796703, + 1534.41992812871, + 1793.515835082672, + 2077.0380524461475, + 2376.259366670552, + 2685.938117275419, + 3002.840663062175, + 3324.9021055377766, + 3650.75738181033, + 3979.474638195215, + 4310.39914822265, + 4643.05898344198 + ] + }, + "hovertemplate": "Method=Matching
Quantiles=%{x}
Quantile loss=%{y}", + "legendgroup": "Matching", + "marker": { + "color": "#117733", + "pattern": { + "shape": "" + } + }, + "name": "Matching", + "offsetgroup": "Matching", + "orientation": "v", + "showlegend": true, + "textposition": "auto", + "type": "bar", + "x": [ + "0.05", + "0.1", + "0.15", + "0.2", + "0.25", + "0.3", + "0.35", + "0.4", + "0.45", + "0.5", + "0.55", + "0.6", + "0.65", + "0.7", + "0.75", + "0.8", + "0.85", + "0.9", + "0.95" + ], + "xaxis": "x", + "y": [ + 101408.0160260291, + 101571.7732345491, + 101735.5304430691, + 101899.2876515891, + 102063.0448601091, + 102226.80206862911, + 102390.5592771491, + 102554.31648566911, + 102718.07369418911, + 102881.83090270912, + 103045.58811122912, + 103209.34531974913, + 103373.10252826913, + 103536.85973678913, + 103700.61694530913, + 103864.37415382915, + 104028.13136234913, + 104191.88857086914, + 104355.64577938915 + ], + "yaxis": "y" + } + ], + "layout": { + "barmode": "group", + "height": 600, + "legend": { + "title": { + "text": "Method" + }, + "tracegroupgap": 0 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "line": { + "color": "#88CCEE", + "dash": "dot", + "width": 2 + }, + "name": "QRF Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 60952.489182045254, + "y1": 60952.489182045254 + }, + { + "line": { + "color": "#CC6677", + "dash": "dot", + "width": 2 + }, + "name": "OLS Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 58409.52427605975, + "y1": 58409.52427605975 + }, + { + "line": { + "color": "#DDCC77", + "dash": "dot", + "width": 2 + }, + "name": "QuantReg Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 54897.71902635497, + "y1": 54897.71902635497 + }, + { + "line": { + "color": "#117733", + "dash": "dot", + "width": 2 + }, + "name": "Matching Mean", + "type": "line", + "x0": -0.5, + "x1": 18.5, + "y0": 102881.83090270912, + "y1": 102881.83090270912 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "House Sales Dataset Benchmarking Results" + }, + "width": 750, + "xaxis": { + "anchor": "y", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": false, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantiles" + }, + "zeroline": false + }, + "yaxis": { + "anchor": "x", + "domain": [ + 0, + 1 + ], + "gridcolor": "#E5E5E5", + "gridwidth": 1, + "linecolor": "#CCCCCC", + "showgrid": true, + "showline": true, + "title": { + "font": { + "size": 12 + }, + "text": "Quantile loss" + }, + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "house_result.cv_results.pop('MDN', None)\n", + "\n", + "print(\"\\n--- Autoimpute CV Results ---\")\n", + "comparison_viz = method_comparison_results(\n", + " data=house_result.cv_results,\n", + " metric=\"quantile_loss\",\n", + ")\n", + "fig = comparison_viz.plot(\n", + " title=\"House Sales Dataset Benchmarking Results\",\n", + " show_mean=True,\n", + ")\n", + "fig.show()" + ] + }, + { + "cell_type": "code", + "execution_count": 46, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "--- Distribution Comparison: Imputed vs Ground Truth (House Sales) ---\n", + "QuantReg: Wasserstein Distance = 64725.2760\n", + "QRF: Wasserstein Distance = 12576.6113\n", + "OLS: Wasserstein Distance = 41862.2873\n", + "Matching: Wasserstein Distance = 8797.8007\n", + "MDN: Wasserstein Distance = 544730.3917\n" + ] + }, + { + "data": { + "text/html": [ + "
\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
VariableMetricDistance
0pricewasserstein_distance64725.276019
\n", + "
" + ], + "text/plain": [ + " Variable Metric Distance\n", + "0 price wasserstein_distance 64725.276019" + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "print(\"\\n--- Distribution Comparison: Imputed vs Ground Truth (House Sales) ---\")\n", + "# Store Wasserstein distances for all methods\n", + "house_wasserstein = {}\n", + "\n", + "distribution_comparison_house = compare_distributions(\n", + " donor_data=house_receiver, # Ground truth test split\n", + " receiver_data=house_result.receiver_data, # Imputed values\n", + " imputed_variables=[house_target],\n", + ")\n", + "best_method_name = house_result.fitted_models[\"best_method\"].__class__.__name__ \n", + "best_method_name = best_method_name.replace(\"Results\", \"\")\n", + "house_wasserstein[best_method_name] = distribution_comparison_house[\n", + " distribution_comparison_house['Metric'] == 'wasserstein_distance'\n", + "]['Distance'].values[0]\n", + "print(f\"{best_method_name}: Wasserstein Distance = {house_wasserstein.get(best_method_name, 'N/A'):.4f}\")\n", + " \n", + "for method_name, imputations in house_result.imputations.items():\n", + " # Skip the 'best_method' (it's a duplicate)\n", + " if method_name == 'best_method': \n", + " continue\n", + " \n", + " # Create a copy of receiver data with this method's imputations \n", + " receiver_with_imputations = house_receiver_no_target.copy() \n", + " \n", + " # Handle both dict (quantile->DataFrame) and DataFrame formats \n", + " if isinstance(imputations, dict): \n", + " # Get median quantile (0.5) imputations \n", + " imp_df = imputations.get(0.5, list(imputations.values())[0]) \n", + " else: \n", + " imp_df = imputations\n", + " \n", + " # Add imputed values \n", + " for var in [house_target]:\n", + " if var in imp_df.columns: \n", + " receiver_with_imputations[var] = imp_df[var].values \n", + " \n", + " # Calculate distribution comparison \n", + " dist_comparison = compare_distributions( \n", + " donor_data=house_receiver, # Ground truth \n", + " receiver_data=receiver_with_imputations,\n", + " imputed_variables=[house_target], \n", + " )\n", + "\n", + " # Extract Wasserstein distance \n", + " wd = dist_comparison[dist_comparison['Metric'] == 'wasserstein_distance']['Distance'].values \n", + " house_wasserstein[method_name] = wd[0]\n", + " \n", + " print(f\"{method_name}: Wasserstein Distance = {house_wasserstein.get(method_name, 'N/A'):.4f}\")\n", + " \n", + "# Display full comparison for best method\n", + "display(distribution_comparison_house)" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "### CIA Sensitivity Analysis\n", + "\n", + "Measuring sensitivity to the Conditional Independence Assumption by progressively removing predictors." + ] + }, + { + "cell_type": "code", + "execution_count": 47, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Running CIA sensitivity analysis...\n" + ] + }, + { + "data": { + "application/vnd.jupyter.widget-view+json": { + "model_id": "a5c135c318a642e389e2e575613cb093", + "version_major": 2, + "version_minor": 0 + }, + "text/plain": [ + "Progressive exclusion: 0%| | 0/10 [00:00\n", + "\n", + "\n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + " \n", + "
DatasetBest MethodQRF RankQRF LossOLS RankOLS LossQuantReg RankQuantReg LossMatching RankMatching LossMDN RankMDN LossQRF CIAOLS CIAQuantReg CIAMatching CIAMDN CIA
0space_gaOLS2.00.0389261.0000000.0361675.0000000.0840554.0000000.0627583.0000000.0504111.4425471.1918640.6876911.2671561.312439
1elevatorsQuantReg3.00.0013212.0000000.0012831.0000000.0011814.0000000.0026345.000000NaN1.2928811.1802571.1739171.0503902.055043
2brazilian_housesQRF1.052.1646673.000000126.9267672.000000104.9360074.000000398.7283715.000000NaN14.6919795.6016207.2676523.0596471.004445
3onlinenewspopularityQuantReg3.01357.2102404.0000001800.6696621.0000001045.3193345.0000001984.7382312.0000001227.6932081.0197691.0027921.0064571.0014641.006107
4abaloneQuantReg3.00.0090562.0000000.0074001.0000000.0064794.0000000.0151455.0000000.0257592.0712492.0669122.1151891.8856061.471810
5house_salesQuantReg3.060952.4891822.00000058409.5242761.00000054897.7190264.000000102881.8309035.000000NaN1.4599851.2979311.3263811.2770180.982165
6Mean Rank-2.510393.6522322.33333310056.1942591.8333339341.3443474.16666717544.2296744.166667409.2564593.6630682.0568962.2628811.5902141.305335
\n", + "" + ], + "text/plain": [ + " Dataset Best Method QRF Rank QRF Loss OLS Rank \\\n", + "0 space_ga OLS 2.0 0.038926 1.000000 \n", + "1 elevators QuantReg 3.0 0.001321 2.000000 \n", + "2 brazilian_houses QRF 1.0 52.164667 3.000000 \n", + "3 onlinenewspopularity QuantReg 3.0 1357.210240 4.000000 \n", + "4 abalone QuantReg 3.0 0.009056 2.000000 \n", + "5 house_sales QuantReg 3.0 60952.489182 2.000000 \n", + "6 Mean Rank - 2.5 10393.652232 2.333333 \n", + "\n", + " OLS Loss QuantReg Rank QuantReg Loss Matching Rank Matching Loss \\\n", + "0 0.036167 5.000000 0.084055 4.000000 0.062758 \n", + "1 0.001283 1.000000 0.001181 4.000000 0.002634 \n", + "2 126.926767 2.000000 104.936007 4.000000 398.728371 \n", + "3 1800.669662 1.000000 1045.319334 5.000000 1984.738231 \n", + "4 0.007400 1.000000 0.006479 4.000000 0.015145 \n", + "5 58409.524276 1.000000 54897.719026 4.000000 102881.830903 \n", + "6 10056.194259 1.833333 9341.344347 4.166667 17544.229674 \n", + "\n", + " MDN Rank MDN Loss QRF CIA OLS CIA QuantReg CIA Matching CIA \\\n", + "0 3.000000 0.050411 1.442547 1.191864 0.687691 1.267156 \n", + "1 5.000000 NaN 1.292881 1.180257 1.173917 1.050390 \n", + "2 5.000000 NaN 14.691979 5.601620 7.267652 3.059647 \n", + "3 2.000000 1227.693208 1.019769 1.002792 1.006457 1.001464 \n", + "4 5.000000 0.025759 2.071249 2.066912 2.115189 1.885606 \n", + "5 5.000000 NaN 1.459985 1.297931 1.326381 1.277018 \n", + "6 4.166667 409.256459 3.663068 2.056896 2.262881 1.590214 \n", + "\n", + " MDN CIA \n", + "0 1.312439 \n", + "1 2.055043 \n", + "2 1.004445 \n", + "3 1.006107 \n", + "4 1.471810 \n", + "5 0.982165 \n", + "6 1.305335 " + ] + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# Collect all CV results\n", + "all_cv_results = {\n", + " \"space_ga\": space_ga_result.cv_results,\n", + " \"elevators\": elevators_result.cv_results,\n", + " \"brazilian_houses\": brazilian_result.cv_results,\n", + " \"onlinenewspopularity\": news_result.cv_results,\n", + " \"abalone\": abalone_result.cv_results,\n", + " \"house_sales\": house_result.cv_results,\n", + "}\n", + "all_cv_results_df = pd.DataFrame(all_cv_results).T\n", + "all_cv_results_df.to_csv(\"benchmark_cv_results.csv\")\n", + "\n", + "# Collect all Wasserstein results \n", + "all_wasserstein_results = {\n", + " \"space_ga\": space_ga_wasserstein, \n", + " \"elevators\": elevators_wasserstein, \n", + " \"brazilian_houses\": brazilian_wasserstein, \n", + " \"onlinenewspopularity\": news_wasserstein, \n", + " \"abalone\": abalone_wasserstein, \n", + " \"house_sales\": house_wasserstein, \n", + "} \n", + "all_wasserstein_results_df = pd.DataFrame(all_wasserstein_results).T\n", + "all_wasserstein_results_df.to_csv(\"benchmark_wasserstein_results.csv\")\n", + "\n", + "# Collect all CIA results\n", + "all_cia_results = {\n", + " \"space_ga\": space_ga_cia_results,\n", + " \"elevators\": elevators_cia_results,\n", + " \"brazilian_houses\": brazilian_cia_results,\n", + " \"onlinenewspopularity\": news_cia_results,\n", + " \"abalone\": abalone_cia_results,\n", + " \"house_sales\": house_cia_results,\n", + "}\n", + "all_cia_results_df = pd.DataFrame(all_cia_results).T\n", + "all_cia_results_df.to_csv(\"benchmark_cia_results.csv\")\n", + "\n", + "# Create summary table\n", + "print(\"\\n=== Cross-Dataset Benchmark Summary Table ===\\n\")\n", + "summary_table = create_benchmark_summary_table(all_cv_results, all_cia_results)\n", + "display(summary_table)" + ] + }, + { + "cell_type": "code", + "execution_count": 2, + "metadata": {}, + "outputs": [], + "source": [ + "all_cv_results = pd.read_csv(\"benchmark_cv_results.csv\", index_col=0).to_dict()\n", + "all_wasserstein_results = pd.read_csv(\"benchmark_wasserstein_results.csv\", index_col=0).to_dict()\n", + "all_cia_results = pd.read_csv(\"benchmark_cia_results.csv\", index_col=0).to_dict()" + ] + }, + { + "cell_type": "code", + "execution_count": 4, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "\n", + "=== Cross-Dataset Benchmark Heatmap ===\n", + "\n" + ] + }, + { + "data": { + "application/vnd.plotly.v1+json": { + "config": { + "plotlyServerURL": "https://plot.ly" + }, + "data": [ + { + "marker": { + "color": "#88CCEE", + "size": 12 + }, + "mode": "markers", + "name": "QRF", + "type": "scatter", + "x": [ + null + ], + "y": [ + null + ] + }, + { + "marker": { + "color": "#CC6677", + "size": 12 + }, + "mode": "markers", + "name": "OLS", + "type": "scatter", + "x": [ + null + ], + "y": [ + null + ] + }, + { + "marker": { + "color": "#DDCC77", + "size": 12 + }, + "mode": "markers", + "name": "QuantReg", + "type": "scatter", + "x": [ + null + ], + "y": [ + null + ] + }, + { + "marker": { + "color": "#117733", + "size": 12 + }, + "mode": "markers", + "name": "Matching", + "type": "scatter", + "x": [ + null + ], + "y": [ + null + ] + }, + { + "marker": { + "color": "#332288", + "size": 12 + }, + "mode": "markers", + "name": "MDN", + "type": "scatter", + "x": [ + null + ], + "y": [ + null + ] + } + ], + "layout": { + "annotations": [ + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0389", + "x": 0, + "xanchor": "center", + "y": 0, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0362", + "x": 1, + "xanchor": "center", + "y": 0, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0841", + "x": 2, + "xanchor": "center", + "y": 0, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0628", + "x": 3, + "xanchor": "center", + "y": 0, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "0.0504", + "x": 4, + "xanchor": "center", + "y": 0, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0101", + "x": 0, + "xanchor": "center", + "y": 1, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0272", + "x": 1, + "xanchor": "center", + "y": 1, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0981", + "x": 2, + "xanchor": "center", + "y": 1, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0111", + "x": 3, + "xanchor": "center", + "y": 1, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "0.0504", + "x": 4, + "xanchor": "center", + "y": 1, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0013", + "x": 0, + "xanchor": "center", + "y": 2, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0013", + "x": 1, + "xanchor": "center", + "y": 2, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0012", + "x": 2, + "xanchor": "center", + "y": 2, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0026", + "x": 3, + "xanchor": "center", + "y": 2, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "N/A", + "x": 4, + "xanchor": "center", + "y": 2, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0003", + "x": 0, + "xanchor": "center", + "y": 3, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0014", + "x": 1, + "xanchor": "center", + "y": 3, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0015", + "x": 2, + "xanchor": "center", + "y": 3, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0002", + "x": 3, + "xanchor": "center", + "y": 3, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "0.0200", + "x": 4, + "xanchor": "center", + "y": 3, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "52.1647", + "x": 0, + "xanchor": "center", + "y": 4, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "126.9268", + "x": 1, + "xanchor": "center", + "y": 4, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "104.9360", + "x": 2, + "xanchor": "center", + "y": 4, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "398.7284", + "x": 3, + "xanchor": "center", + "y": 4, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "N/A", + "x": 4, + "xanchor": "center", + "y": 4, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "27.1747", + "x": 0, + "xanchor": "center", + "y": 5, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "99.3468", + "x": 1, + "xanchor": "center", + "y": 5, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "112.3643", + "x": 2, + "xanchor": "center", + "y": 5, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "120.4604", + "x": 3, + "xanchor": "center", + "y": 5, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "3378.6172", + "x": 4, + "xanchor": "center", + "y": 5, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "1357.2102", + "x": 0, + "xanchor": "center", + "y": 6, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "1800.6697", + "x": 1, + "xanchor": "center", + "y": 6, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "1045.3193", + "x": 2, + "xanchor": "center", + "y": 6, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "1984.7382", + "x": 3, + "xanchor": "center", + "y": 6, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "1227.6932", + "x": 4, + "xanchor": "center", + "y": 6, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "783.7171", + "x": 0, + "xanchor": "center", + "y": 7, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "2777.1152", + "x": 1, + "xanchor": "center", + "y": 7, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "2376.8148", + "x": 2, + "xanchor": "center", + "y": 7, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "239.8596", + "x": 3, + "xanchor": "center", + "y": 7, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "2038.0328", + "x": 4, + "xanchor": "center", + "y": 7, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0091", + "x": 0, + "xanchor": "center", + "y": 8, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0074", + "x": 1, + "xanchor": "center", + "y": 8, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0065", + "x": 2, + "xanchor": "center", + "y": 8, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0151", + "x": 3, + "xanchor": "center", + "y": 8, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "0.0258", + "x": 4, + "xanchor": "center", + "y": 8, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0028", + "x": 0, + "xanchor": "center", + "y": 9, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0037", + "x": 1, + "xanchor": "center", + "y": 9, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "0.0027", + "x": 2, + "xanchor": "center", + "y": 9, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "0.0025", + "x": 3, + "xanchor": "center", + "y": 9, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "0.0404", + "x": 4, + "xanchor": "center", + "y": 9, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "60952.4892", + "x": 0, + "xanchor": "center", + "y": 10, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "58409.5243", + "x": 1, + "xanchor": "center", + "y": 10, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "54897.7190", + "x": 2, + "xanchor": "center", + "y": 10, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "102881.8309", + "x": 3, + "xanchor": "center", + "y": 10, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "N/A", + "x": 4, + "xanchor": "center", + "y": 10, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "12576.6113", + "x": 0, + "xanchor": "center", + "y": 11, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "41862.2873", + "x": 1, + "xanchor": "center", + "y": 11, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 10 + }, + "showarrow": false, + "text": "64725.2760", + "x": 2, + "xanchor": "center", + "y": 11, + "yanchor": "middle" + }, + { + "font": { + "color": "black", + "size": 12 + }, + "showarrow": false, + "text": "8797.8007", + "x": 3, + "xanchor": "center", + "y": 11, + "yanchor": "middle" + }, + { + "font": { + "color": "white", + "size": 10 + }, + "showarrow": false, + "text": "544730.3917", + "x": 4, + "xanchor": "center", + "y": 11, + "yanchor": "middle" + } + ], + "height": 580, + "legend": { + "orientation": "h", + "x": 0.5, + "xanchor": "center", + "y": -0.02, + "yanchor": "top" + }, + "margin": { + "b": 40, + "l": 180, + "r": 20, + "t": 60 + }, + "paper_bgcolor": "#FAFAFA", + "plot_bgcolor": "#FAFAFA", + "shapes": [ + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": -0.5, + "y1": 0.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": -0.5, + "y1": 0.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": -0.5, + "y1": 0.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": -0.5, + "y1": 0.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": -0.5, + "y1": 0.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 0.5, + "y1": 1.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 0.5, + "y1": 1.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 0.5, + "y1": 1.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 0.5, + "y1": 1.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 0.5, + "y1": 1.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 1.5, + "y1": 2.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 1.5, + "y1": 2.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 1.5, + "y1": 2.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 1.5, + "y1": 2.5 + }, + { + "fillcolor": "rgba(200, 200, 200, 0.3)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 1.5, + "y1": 2.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 2.5, + "y1": 3.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 2.5, + "y1": 3.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 2.5, + "y1": 3.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 2.5, + "y1": 3.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 2.5, + "y1": 3.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 3.5, + "y1": 4.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 3.5, + "y1": 4.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 3.5, + "y1": 4.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 3.5, + "y1": 4.5 + }, + { + "fillcolor": "rgba(200, 200, 200, 0.3)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 3.5, + "y1": 4.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 4.5, + "y1": 5.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 4.5, + "y1": 5.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 4.5, + "y1": 5.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 4.5, + "y1": 5.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 4.5, + "y1": 5.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 5.5, + "y1": 6.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 5.5, + "y1": 6.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 5.5, + "y1": 6.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 5.5, + "y1": 6.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 5.5, + "y1": 6.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 6.5, + "y1": 7.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 6.5, + "y1": 7.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 6.5, + "y1": 7.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 6.5, + "y1": 7.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 6.5, + "y1": 7.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 7.5, + "y1": 8.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 7.5, + "y1": 8.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 7.5, + "y1": 8.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 7.5, + "y1": 8.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 7.5, + "y1": 8.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 8.5, + "y1": 9.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 8.5, + "y1": 9.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 8.5, + "y1": 9.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 8.5, + "y1": 9.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 8.5, + "y1": 9.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 9.5, + "y1": 10.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 9.5, + "y1": 10.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 9.5, + "y1": 10.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.80)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 9.5, + "y1": 10.5 + }, + { + "fillcolor": "rgba(200, 200, 200, 0.3)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 9.5, + "y1": 10.5 + }, + { + "fillcolor": "rgba(136, 204, 238, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": -0.5, + "x1": 0.5, + "y0": 10.5, + "y1": 11.5 + }, + { + "fillcolor": "rgba(204, 102, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 0.5, + "x1": 1.5, + "y0": 10.5, + "y1": 11.5 + }, + { + "fillcolor": "rgba(221, 204, 119, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 1.5, + "x1": 2.5, + "y0": 10.5, + "y1": 11.5 + }, + { + "fillcolor": "rgba(17, 119, 51, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 2.5, + "x1": 3.5, + "y0": 10.5, + "y1": 11.5 + }, + { + "fillcolor": "rgba(51, 34, 136, 0.50)", + "line": { + "color": "white", + "width": 1 + }, + "type": "rect", + "x0": 3.5, + "x1": 4.5, + "y0": 10.5, + "y1": 11.5 + } + ], + "template": { + "data": { + "bar": [ + { + "error_x": { + "color": "#2a3f5f" + }, + "error_y": { + "color": "#2a3f5f" + }, + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "bar" + } + ], + "barpolar": [ + { + "marker": { + "line": { + "color": "#E5ECF6", + "width": 0.5 + }, + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "barpolar" + } + ], + "carpet": [ + { + "aaxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "baxis": { + "endlinecolor": "#2a3f5f", + "gridcolor": "white", + "linecolor": "white", + "minorgridcolor": "white", + "startlinecolor": "#2a3f5f" + }, + "type": "carpet" + } + ], + "choropleth": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "choropleth" + } + ], + "contour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "contour" + } + ], + "contourcarpet": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "contourcarpet" + } + ], + "heatmap": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmap" + } + ], + "heatmapgl": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "heatmapgl" + } + ], + "histogram": [ + { + "marker": { + "pattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + } + }, + "type": "histogram" + } + ], + "histogram2d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2d" + } + ], + "histogram2dcontour": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "histogram2dcontour" + } + ], + "mesh3d": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "type": "mesh3d" + } + ], + "parcoords": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "parcoords" + } + ], + "pie": [ + { + "automargin": true, + "type": "pie" + } + ], + "scatter": [ + { + "fillpattern": { + "fillmode": "overlay", + "size": 10, + "solidity": 0.2 + }, + "type": "scatter" + } + ], + "scatter3d": [ + { + "line": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatter3d" + } + ], + "scattercarpet": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattercarpet" + } + ], + "scattergeo": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergeo" + } + ], + "scattergl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattergl" + } + ], + "scattermapbox": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scattermapbox" + } + ], + "scatterpolar": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolar" + } + ], + "scatterpolargl": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterpolargl" + } + ], + "scatterternary": [ + { + "marker": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "type": "scatterternary" + } + ], + "surface": [ + { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + }, + "colorscale": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "type": "surface" + } + ], + "table": [ + { + "cells": { + "fill": { + "color": "#EBF0F8" + }, + "line": { + "color": "white" + } + }, + "header": { + "fill": { + "color": "#C8D4E3" + }, + "line": { + "color": "white" + } + }, + "type": "table" + } + ] + }, + "layout": { + "annotationdefaults": { + "arrowcolor": "#2a3f5f", + "arrowhead": 0, + "arrowwidth": 1 + }, + "autotypenumbers": "strict", + "coloraxis": { + "colorbar": { + "outlinewidth": 0, + "ticks": "" + } + }, + "colorscale": { + "diverging": [ + [ + 0, + "#8e0152" + ], + [ + 0.1, + "#c51b7d" + ], + [ + 0.2, + "#de77ae" + ], + [ + 0.3, + "#f1b6da" + ], + [ + 0.4, + "#fde0ef" + ], + [ + 0.5, + "#f7f7f7" + ], + [ + 0.6, + "#e6f5d0" + ], + [ + 0.7, + "#b8e186" + ], + [ + 0.8, + "#7fbc41" + ], + [ + 0.9, + "#4d9221" + ], + [ + 1, + "#276419" + ] + ], + "sequential": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ], + "sequentialminus": [ + [ + 0, + "#0d0887" + ], + [ + 0.1111111111111111, + "#46039f" + ], + [ + 0.2222222222222222, + "#7201a8" + ], + [ + 0.3333333333333333, + "#9c179e" + ], + [ + 0.4444444444444444, + "#bd3786" + ], + [ + 0.5555555555555556, + "#d8576b" + ], + [ + 0.6666666666666666, + "#ed7953" + ], + [ + 0.7777777777777778, + "#fb9f3a" + ], + [ + 0.8888888888888888, + "#fdca26" + ], + [ + 1, + "#f0f921" + ] + ] + }, + "colorway": [ + "#636efa", + "#EF553B", + "#00cc96", + "#ab63fa", + "#FFA15A", + "#19d3f3", + "#FF6692", + "#B6E880", + "#FF97FF", + "#FECB52" + ], + "font": { + "color": "#2a3f5f" + }, + "geo": { + "bgcolor": "white", + "lakecolor": "white", + "landcolor": "#E5ECF6", + "showlakes": true, + "showland": true, + "subunitcolor": "white" + }, + "hoverlabel": { + "align": "left" + }, + "hovermode": "closest", + "mapbox": { + "style": "light" + }, + "paper_bgcolor": "white", + "plot_bgcolor": "#E5ECF6", + "polar": { + "angularaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "radialaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "scene": { + "xaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "yaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + }, + "zaxis": { + "backgroundcolor": "#E5ECF6", + "gridcolor": "white", + "gridwidth": 2, + "linecolor": "white", + "showbackground": true, + "ticks": "", + "zerolinecolor": "white" + } + }, + "shapedefaults": { + "line": { + "color": "#2a3f5f" + } + }, + "ternary": { + "aaxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "baxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + }, + "bgcolor": "#E5ECF6", + "caxis": { + "gridcolor": "white", + "linecolor": "white", + "ticks": "" + } + }, + "title": { + "x": 0.05 + }, + "xaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + }, + "yaxis": { + "automargin": true, + "gridcolor": "white", + "linecolor": "white", + "ticks": "", + "title": { + "standoff": 15 + }, + "zerolinecolor": "white", + "zerolinewidth": 2 + } + } + }, + "title": { + "font": { + "size": 14 + }, + "text": "Cross-Dataset Performance Comparison" + }, + "width": 700, + "xaxis": { + "range": [ + -0.5, + 4.5 + ], + "showgrid": false, + "side": "top", + "tickmode": "array", + "ticktext": [ + "QRF", + "OLS", + "QuantReg", + "Matching", + "MDN" + ], + "tickvals": [ + 0, + 1, + 2, + 3, + 4 + ], + "zeroline": false + }, + "yaxis": { + "autorange": "reversed", + "range": [ + -0.5, + 11.5 + ], + "showgrid": false, + "tickmode": "array", + "ticktext": [ + "space_ga (Q-Loss)", + "space_ga (W-Dist)", + "elevators (Q-Loss)", + "elevators (W-Dist)", + "brazilian_houses (Q-Loss)", + "brazilian_houses (W-Dist)", + "onlinenewspopularity (Q-Loss)", + "onlinenewspopularity (W-Dist)", + "abalone (Q-Loss)", + "abalone (W-Dist)", + "house_sales (Q-Loss)", + "house_sales (W-Dist)" + ], + "tickvals": [ + 0, + 1, + 2, + 3, + 4, + 5, + 6, + 7, + 8, + 9, + 10, + 11 + ], + "zeroline": false + } + } + } + }, + "metadata": {}, + "output_type": "display_data" + } + ], + "source": [ + "# Create benchmark heatmap\n", + "print(\"\\n=== Cross-Dataset Benchmark Heatmap ===\\n\")\n", + "heatmap_fig = create_benchmark_heatmap(\n", + " all_cv_results,\n", + " wasserstein_results=all_wasserstein_results,\n", + " title=\"Cross-Dataset Performance Comparison\"\n", + ")\n", + "heatmap_fig.show()" + ] + }, + { + "cell_type": "markdown", + "metadata": {}, + "source": [ + "---\n", + "## Summary\n", + "\n", + "This notebook has analyzed 6 benchmarking datasets:\n", + "\n", + "1. **space_ga** - Georgia voting data (3,107 rows, 7 cols)\n", + "2. **elevators** - Aircraft control data (16,599 rows, 19 cols)\n", + "3. **brazilian_houses** - Real estate data (10,692 rows, 13 cols)\n", + "4. **onlinenewspopularity** - News article data (39,644 rows, 60 cols)\n", + "5. **abalone** - Marine biology data (4,177 rows, 9 cols)\n", + "6. **house_sales** - King County real estate (21,613 rows, 22 cols)\n", + "\n", + "For each dataset:\n", + "- Computed predictor correlations (Pearson, Spearman, Mutual Information)\n", + "- Performed leave-one-out analysis to identify important predictors\n", + "- Split data into donor/receiver sets (60/40)\n", + "- Ran autoimpute to compare QRF, OLS, QuantReg, Matching, and MDN models\n", + "- Compared imputed vs ground truth distributions" + ] + } + ], + "metadata": { + "kernelspec": { + "display_name": "pe3.13", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.13.0" + } + }, + "nbformat": 4, + "nbformat_minor": 4 +} diff --git a/tests/test_autoimpute.py b/tests/test_autoimpute.py index c58ef5f..8b2ceb8 100644 --- a/tests/test_autoimpute.py +++ b/tests/test_autoimpute.py @@ -448,4 +448,4 @@ def test_autoimpute_consistency(simple_data: tuple) -> None: "mean_test" ] if not np.isnan(loss1) and not np.isnan(loss2): - np.testing.assert_allclose(loss1, loss2, rtol=0.05) + np.testing.assert_allclose(loss1, loss2, rtol=0.10) diff --git a/tests/test_dashboard_formatter.py b/tests/test_dashboard_formatter.py index 9cc49d4..f42b3d4 100644 --- a/tests/test_dashboard_formatter.py +++ b/tests/test_dashboard_formatter.py @@ -14,7 +14,6 @@ from microimpute.utils.dashboard_formatter import format_csv - # Valid type values that should appear in the output VALID_TYPES = { "benchmark_loss", @@ -34,6 +33,7 @@ "quantile", "metric_name", "metric_value", + "metric_std", "split", "additional_info", ] @@ -57,8 +57,18 @@ def sample_autoimpute_result(): }, index=["train", "test"], ), + "results_std": pd.DataFrame( + { + 0.1: [0.001, 0.002], + 0.5: [0.0015, 0.0025], + 0.9: [0.002, 0.003], + }, + index=["train", "test"], + ), "mean_train": 0.015, "mean_test": 0.025, + "std_train": 0.0015, + "std_test": 0.0025, "variables": ["var1", "var2"], }, "log_loss": { @@ -68,8 +78,16 @@ def sample_autoimpute_result(): }, index=["train", "test"], ), + "results_std": pd.DataFrame( + { + 0.5: [0.01, 0.015], + }, + index=["train", "test"], + ), "mean_train": 0.1, "mean_test": 0.15, + "std_train": 0.01, + "std_test": 0.015, "variables": ["cat_var"], }, }, @@ -83,8 +101,18 @@ def sample_autoimpute_result(): }, index=["train", "test"], ), + "results_std": pd.DataFrame( + { + 0.1: [0.0012, 0.0022], + 0.5: [0.0017, 0.0027], + 0.9: [0.0022, 0.0032], + }, + index=["train", "test"], + ), "mean_train": 0.017, "mean_test": 0.027, + "std_train": 0.0017, + "std_test": 0.0027, "variables": ["var1", "var2"], }, }, @@ -350,6 +378,60 @@ def test_best_method_marked_correctly(self, sample_autoimpute_result): finally: Path(output_path).unlink() + def test_metric_std_column_populated(self, sample_autoimpute_result): + """Test that metric_std column is populated from CV results.""" + with tempfile.NamedTemporaryFile( + mode="w", delete=False, suffix=".csv" + ) as f: + output_path = f.name + + try: + result = format_csv( + output_path=output_path, + autoimpute_result=sample_autoimpute_result, + ) + + # Check that metric_std column exists + assert "metric_std" in result.columns + + benchmark_rows = result[result["type"] == "benchmark_loss"] + + # Check per-quantile rows have std values + ols_quantile_rows = benchmark_rows[ + (benchmark_rows["method"] == "OLS") + & (benchmark_rows["quantile"] != "mean") + & (benchmark_rows["metric_name"] == "quantile_loss") + ] + + # All per-quantile rows should have non-null std + assert not ols_quantile_rows["metric_std"].isna().any() + assert (ols_quantile_rows["metric_std"] >= 0).all() + + # Check mean rows have std values + ols_mean_rows = benchmark_rows[ + (benchmark_rows["method"] == "OLS") + & (benchmark_rows["quantile"] == "mean") + & (benchmark_rows["metric_name"] == "quantile_loss") + ] + + assert not ols_mean_rows["metric_std"].isna().any() + + # Verify specific std value matches fixture + # Test split should have std_test = 0.0025 for OLS quantile_loss + test_mean_row = benchmark_rows[ + (benchmark_rows["method"] == "OLS") + & (benchmark_rows["quantile"] == "mean") + & (benchmark_rows["metric_name"] == "quantile_loss") + & (benchmark_rows["split"] == "test") + ] + assert len(test_mean_row) == 1 + assert np.isclose( + test_mean_row.iloc[0]["metric_std"], 0.0025, rtol=1e-10 + ) + + finally: + Path(output_path).unlink() + class TestFormatCSVDistributionDistance: """Tests for distribution_distance type formatting.""" diff --git a/tests/test_models/test_imputers.py b/tests/test_models/test_imputers.py index 18548b8..762d45d 100644 --- a/tests/test_models/test_imputers.py +++ b/tests/test_models/test_imputers.py @@ -17,7 +17,6 @@ from microimpute.models import * from microimpute.utils.data import preprocess_data - # === Fixtures === diff --git a/tests/test_models/test_mdn.py b/tests/test_models/test_mdn.py index ea74842..9e76bc6 100644 --- a/tests/test_models/test_mdn.py +++ b/tests/test_models/test_mdn.py @@ -20,7 +20,6 @@ _generate_data_hash, ) - # === Fixtures === diff --git a/tests/test_models/test_ols.py b/tests/test_models/test_ols.py index 94adfbb..4db9686 100644 --- a/tests/test_models/test_ols.py +++ b/tests/test_models/test_ols.py @@ -13,7 +13,6 @@ from microimpute.utils.data import preprocess_data from microimpute.visualizations import * - # === Fixtures === diff --git a/tests/test_models/test_qrf.py b/tests/test_models/test_qrf.py index 2f26c52..04d740a 100644 --- a/tests/test_models/test_qrf.py +++ b/tests/test_models/test_qrf.py @@ -21,7 +21,6 @@ from microimpute.utils.data import preprocess_data from microimpute.visualizations import * - # === Fixtures and Test Data === diff --git a/tests/test_models/test_quantreg.py b/tests/test_models/test_quantreg.py index 5caee74..74e241f 100644 --- a/tests/test_models/test_quantreg.py +++ b/tests/test_models/test_quantreg.py @@ -13,7 +13,6 @@ from microimpute.utils.data import preprocess_data from microimpute.visualizations import * - # === Fixtures === diff --git a/tests/test_visualizations.py b/tests/test_visualizations.py index df0e5c3..ad1475d 100644 --- a/tests/test_visualizations.py +++ b/tests/test_visualizations.py @@ -661,3 +661,124 @@ def test_nan_handling(self): # Should handle NaNs gracefully fig = viz.plot() assert fig is not None + + +class TestErrorBars: + """Test error bar functionality in visualizations.""" + + def test_performance_results_with_error_bars(self): + """Test PerformanceResults displays error bars when std data is present.""" + np.random.seed(42) + quantiles = [0.1, 0.5, 0.9] + + results_df = pd.DataFrame( + {q: np.random.uniform(0.1, 0.5, 2) for q in quantiles}, + index=["train", "test"], + ) + std_df = pd.DataFrame( + {q: np.random.uniform(0.01, 0.05, 2) for q in quantiles}, + index=["train", "test"], + ) + + results = { + "quantile_loss": { + "results": results_df, + "results_std": std_df, + "mean_train": results_df.loc["train"].mean(), + "mean_test": results_df.loc["test"].mean(), + "std_train": std_df.loc["train"].mean(), + "std_test": std_df.loc["test"].mean(), + "variables": ["var1"], + } + } + + with warnings.catch_warnings(): + warnings.simplefilter("ignore") + + viz = PerformanceResults( + results=results, + metric="quantile_loss", + model_name="TestModel", + ) + + fig = viz.plot(show_error_bars=True) + assert fig is not None + + # Check that error bars are present + has_error_bars = any( + trace.error_y is not None and trace.error_y.array is not None + for trace in fig.data + ) + assert has_error_bars + + def test_comparison_results_with_error_bars(self): + """Test MethodComparisonResults displays error bars when std is present.""" + np.random.seed(42) + quantiles = [0.1, 0.5, 0.9] + + results_df = pd.DataFrame( + {q: np.random.uniform(0.1, 0.5, 2) for q in quantiles}, + index=["train", "test"], + ) + std_df = pd.DataFrame( + {q: np.random.uniform(0.01, 0.05, 2) for q in quantiles}, + index=["train", "test"], + ) + + comparison_results = { + "OLS": { + "quantile_loss": { + "results": results_df, + "results_std": std_df, + "mean_test": results_df.loc["test"].mean(), + "std_test": std_df.loc["test"].mean(), + "variables": ["var1"], + } + }, + "QRF": { + "quantile_loss": { + "results": results_df * 0.9, + "results_std": std_df * 1.1, + "mean_test": (results_df * 0.9).loc["test"].mean(), + "std_test": (std_df * 1.1).loc["test"].mean(), + "variables": ["var1"], + } + }, + } + + with warnings.catch_warnings(): + warnings.simplefilter("ignore") + + viz = MethodComparisonResults( + comparison_results, metric="quantile_loss" + ) + + fig = viz.plot(show_error_bars=True) + assert fig is not None + + # Check that Std column is in the data + assert "Std" in viz.comparison_data.columns + assert not viz.comparison_data["Std"].isna().all() + + def test_cv_results_contain_std(self, diabetes_data): + """Test that cross_validate_model returns std information.""" + predictors = ["age", "sex", "bmi", "bp"] + imputed_variables = ["s1"] + + data = diabetes_data[predictors + imputed_variables] + data = preprocess_data(data, full_data=True) + + with warnings.catch_warnings(): + warnings.simplefilter("ignore") + + results = cross_validate_model( + OLS, data, predictors, imputed_variables, n_splits=3 + ) + + ql = results["quantile_loss"] + assert "results_std" in ql + assert "std_train" in ql + assert "std_test" in ql + assert ql["std_train"] >= 0 + assert ql["std_test"] >= 0 + assert (ql["results_std"] >= 0).all().all()