models.py

from sklearn.naive_bayes import GaussianNB
from sklearn.tree import DecisionTreeClassifier
from sklearn.neural_network import MLPClassifier
from sklearn.model_selection import GridSearchCV


def train_naive_bayes(X, y):
    model = GaussianNB()
    model.fit(X, y)
    return model

def train_decision_tree(X, y, max_depth=None):
    model = DecisionTreeClassifier(
        max_depth=max_depth, random_state=42
    )
    model.fit(X, y)
    return model

def train_mlp(X, y,
              hidden_layers=(128,),
              max_iter=300,
              alpha=1e-4,
              learning_rate_init=1e-3):
    model = MLPClassifier(
        hidden_layer_sizes=hidden_layers,
        activation='relu',
        solver='adam',
        alpha=alpha,
        batch_size='auto',
        learning_rate='adaptive',
        learning_rate_init=learning_rate_init,
        early_stopping=True,          # stop if no improvement on a 10% val-split
        validation_fraction=0.1,
        n_iter_no_change=10,
        max_iter=max_iter,
        random_state=42
    )
def tune_nb(X, y):
    grid = GridSearchCV(
        GaussianNB(),
        {"var_smoothing": [1e-9, 1e-8, 1e-7, 1e-6]},
        cv=3, scoring="accuracy", n_jobs=-1
    )
    grid.fit(X, y)
    return grid.best_estimator_

def tune_dt(X, y):
    params = {
        "max_depth": [5, 10, 15, None],
        "min_samples_leaf": [1, 5, 10]
    }
    grid = GridSearchCV(
        DecisionTreeClassifier(random_state=42),
        params, cv=3, scoring="accuracy", n_jobs=-1
    )
    grid.fit(X, y)
    return grid.best_estimator_

def tune_mlp(X, y):
    base = MLPClassifier(
        solver="adam", early_stopping=True,
        validation_fraction=0.1, n_iter_no_change=10,
        max_iter=300, random_state=42
    )
    grid = GridSearchCV(
        base,
        {
          "hidden_layer_sizes": [(128,), (256,128)],
          "alpha": [1e-4, 1e-3],
          "learning_rate_init": [1e-3, 5e-4]
        },
        cv=3, scoring="accuracy", n_jobs=-1
    )

    grid.fit(X, y)
    return grid.best_estimator_