feat(proportion.single.inequality): add exact test by Snoopy1866 · Pull Request #334 · Snoopy1866/pystatpower

Snoopy1866 · 2026-05-11T07:03:20Z

No description provided.

read-the-docs-community · 2026-05-11T07:03:48Z

Documentation build overview

📚 pystatpower | 🛠️ Build #33052998 | 📁 Comparing bef034c against latest (a1af8d3)

🔍 Preview build

33 files changed · + 16 added · ± 17 modified

+ Added

± Modified

gemini-code-assist

Code Review

This pull request introduces exact binomial test power and sample size calculations for one-sample proportion tests. The feedback identifies several critical issues: the signature change in the _power function breaks existing calls in other modules, and the solve_size implementation for exact tests is incomplete, lacking both a return statement and the logic to determine the minimum sample size. Furthermore, the power calculations for lower and two-sided tails are currently anti-conservative and require adjustments to the critical values. Finally, the documentation for the new exact test feature needs to be populated with mathematical background.

gemini-code-assist · 2026-05-11T07:05:20Z

    alpha: float,
    phat: bool,
    continuity_correction: bool,
+    exact: bool,


Adding exact as a required positional argument to _power breaks existing calls in solve_null_proportion and solve_proportion, which currently only pass 7 arguments. Consider providing a default value for exact to maintain backward compatibility and fix the TypeError in those functions.

Suggested change

exact: bool,

exact: bool = False,

gemini-code-assist · 2026-05-11T07:05:20Z

+    if exact:
+        initial_size = solve_size(null_proportion, proportion, alternative, alpha, power, exact=False)
+        search_interval_lower_bound = max(floor(0.5 * initial_size), 1)
+        search_interval_upper_bound = min(ceil(initial_size * 1.5), SAMPLE_SIZE_SEARCH_MAX)

-    return ceil(brentq(func, 0.000001, SAMPLE_SIZE_SEARCH_MAX))
+        rng = range(search_interval_lower_bound, search_interval_upper_bound)
+        dict_power = dict.fromkeys(rng)
+        for size in rng:
+            dict_power[size] = solve_power(null_proportion, proportion, size, alternative, alpha, exact)
+    else:
+        return ceil(brentq(func, 0.000001, SAMPLE_SIZE_SEARCH_MAX))


The implementation for exact=True in solve_size is incomplete and contains a critical bug: it lacks a return statement, causing the function to return None. Additionally, the current logic populates a dictionary but does not use it to find the minimum sample size. The search logic should iterate through the range and return the smallest size that achieves the target power. Also, ensure that exact is passed correctly to solve_power (it is currently being passed as the 6th positional argument, which corresponds to phat).

Suggested change

if exact:

initial_size = solve_size(null_proportion, proportion, alternative, alpha, power, exact=False)

search_interval_lower_bound = max(floor(0.5 * initial_size), 1)

search_interval_upper_bound = min(ceil(initial_size * 1.5), SAMPLE_SIZE_SEARCH_MAX)

return ceil(brentq(func, 0.000001, SAMPLE_SIZE_SEARCH_MAX))

rng = range(search_interval_lower_bound, search_interval_upper_bound)

dict_power = dict.fromkeys(rng)

for size in rng:

dict_power[size] = solve_power(null_proportion, proportion, size, alternative, alpha, exact)

else:

return ceil(brentq(func, 0.000001, SAMPLE_SIZE_SEARCH_MAX))

if exact:

initial_size = solve_size(null_proportion, proportion, alternative, alpha, power, exact=False)

search_interval_lower_bound = max(floor(0.5 * initial_size), 1)

search_interval_upper_bound = min(ceil(initial_size * 1.5), SAMPLE_SIZE_SEARCH_MAX)

for size in range(search_interval_lower_bound, search_interval_upper_bound + 1):

if solve_power(null_proportion, proportion, size, alternative, alpha, phat, continuity_correction, exact) >= power:

return size

return search_interval_upper_bound

else:

return ceil(brentq(func, 0.000001, SAMPLE_SIZE_SEARCH_MAX))

gemini-code-assist · 2026-05-11T07:05:20Z

+                z_alpha = binom.ppf(alpha, size, null_proportion)
+                power = binom.cdf(z_alpha, size, proportion)


The power calculation for the lower tail is anti-conservative. binom.ppf(alpha, ...) returns the smallest $k$ such that $P(X \le k) \ge \alpha$. If $P(X \le k) > \alpha$, using this $k$ as the critical value results in a test with a significance level greater than $\alpha$. For an exact test, the critical value should be the largest $k$ such that $P(X \le k) \le \alpha$.

Suggested change

z_alpha = binom.ppf(alpha, size, null_proportion)

power = binom.cdf(z_alpha, size, proportion)

z_alpha = binom.ppf(alpha, size, null_proportion)

if binom.cdf(z_alpha, size, null_proportion) > alpha:

z_alpha -= 1

power = binom.cdf(z_alpha, size, proportion)

gemini-code-assist · 2026-05-11T07:05:20Z

+            z_alpha_2 = binom.ppf(alpha / 2, size, null_proportion)
+            power = 1 - binom.cdf(z_alpha_1, size, proportion) + binom.cdf(z_alpha_2, size, proportion)


Similar to the one-sided lower tail, the two-sided test implementation might be anti-conservative because z_alpha_2 (the lower critical value) is calculated using ppf without checking if the resulting CDF exceeds $\alpha/2$.

Suggested change

z_alpha_2 = binom.ppf(alpha / 2, size, null_proportion)

power = 1 - binom.cdf(z_alpha_1, size, proportion) + binom.cdf(z_alpha_2, size, proportion)

z_alpha_2 = binom.ppf(alpha / 2, size, null_proportion)

if binom.cdf(z_alpha_2, size, null_proportion) > alpha / 2:

z_alpha_2 -= 1

power = 1 - binom.cdf(z_alpha_1, size, proportion) + binom.cdf(z_alpha_2, size, proportion)

gemini-code-assist · 2026-05-11T07:05:20Z

    n = \frac{n'}{4} \left( 1 + \sqrt{1 + \frac{2}{|p_1 - p_0| n'}}\right)^2
    $$
+
+## 精确检验 {exact-test}


The "精确检验" (Exact Test) section is currently empty. Please add the mathematical background and formulas for the binomial exact test power calculation to provide complete documentation for this new feature.

codacy-production

Pull Request Overview

This pull request does not contain any code changes or a description, despite the title indicating a new feature implementation for 'proportion.single.inequality'. The acceptance criterion to implement an exact test calculation for single proportion inequality is currently unaddressed. Because no files were modified, the PR is effectively empty and cannot be merged in its current state.

About this PR

The pull request contains no code changes. Please ensure the intended commits are pushed and included in the branch.
The PR description is empty. Please provide context, methodology, and any relevant references for the exact test being implemented.

Test suggestions

The function correctly calculates the p-value for a standard binomial distribution case.
The function handles edge cases where the number of successes is 0 or equal to the total number of trials.
The function returns appropriate errors or values for invalid inputs (e.g., negative counts or proportions outside [0, 1])

Prompt proposal for missing tests

Consider implementing these tests if applicable:
1. The function correctly calculates the p-value for a standard binomial distribution case.
2. The function handles edge cases where the number of successes is 0 or equal to the total number of trials.
3. The function returns appropriate errors or values for invalid inputs (e.g., negative counts or proportions outside [0, 1])

_{TIP Improve review quality by adding custom instructions}
_{TIP How was this review? Give us feedback}

codacy-production · 2026-05-29T01:37:41Z

Not up to standards ⛔

🔴 Issues 4 high

Alerts:
⚠ 4 issues (≤ 0 issues of at least minor severity)

Results:
4 new issues

Category Results

ErrorProne 4 high

View in Codacy

🟢 Metrics 4 complexity

Metric Results

Complexity 4

View in Codacy

AI Reviewer: first review requested successfully. AI can make mistakes. Always validate suggestions.

_{TIP This summary will be updated as you push new changes.}

github-actions Bot added the documentation Improvements or additions to documentation label May 11, 2026

gemini-code-assist Bot reviewed May 11, 2026

View reviewed changes

Snoopy1866 force-pushed the feat-proportion-single-inequality-exact branch 2 times, most recently from b5b7cab to 247b762 Compare May 28, 2026 02:47

codacy-production Bot reviewed May 29, 2026

View reviewed changes

feat(proportion.single.inequality): add exact test

bef034c

Snoopy1866 force-pushed the feat-proportion-single-inequality-exact branch from 247b762 to bef034c Compare June 9, 2026 07:05

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat(proportion.single.inequality): add exact test#334

feat(proportion.single.inequality): add exact test#334
Snoopy1866 wants to merge 1 commit into
mainfrom
feat-proportion-single-inequality-exact

Snoopy1866 commented May 11, 2026

Uh oh!

read-the-docs-community Bot commented May 11, 2026 •

edited

Loading

Uh oh!

gemini-code-assist Bot left a comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

gemini-code-assist Bot May 11, 2026

Uh oh!

codacy-production Bot left a comment

Uh oh!

codacy-production Bot commented May 29, 2026 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

		z_alpha = binom.ppf(alpha, size, null_proportion)
		power = binom.cdf(z_alpha, size, proportion)

		z_alpha_2 = binom.ppf(alpha / 2, size, null_proportion)
		power = 1 - binom.cdf(z_alpha_1, size, proportion) + binom.cdf(z_alpha_2, size, proportion)

Conversation

Snoopy1866 commented May 11, 2026

Uh oh!

read-the-docs-community Bot commented May 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Documentation build overview

Uh oh!

gemini-code-assist Bot left a comment

Choose a reason for hiding this comment

Code Review

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

gemini-code-assist Bot May 11, 2026

Choose a reason for hiding this comment

Uh oh!

codacy-production Bot left a comment

Choose a reason for hiding this comment

Pull Request Overview

About this PR

Test suggestions

Uh oh!

codacy-production Bot commented May 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Not up to standards ⛔

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

1 participant

read-the-docs-community Bot commented May 11, 2026 •

edited

Loading

codacy-production Bot commented May 29, 2026 •

edited

Loading