Fixes #453: _correl_pvalue does not divide by zero if r==1

[AlexanderJCS]

Before this change, _correl_pvalue would create a division by zero warning when r==1.

Simply adding the condition near the top of the function:

    if np.isclose(r**2, 1):  # Avoid divide by zero error
        return 0.0  # Since p value approaches 0 as r approaches 1, just return 0

Fixes this. This assumes that n >= 2 since p would be undefined if n < 2, but I did not account for this case since it seems that the function already assumes that n >= 2.

Please let me know if you want me to make any other changes.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 98.54%. Comparing base (c939d2d) to head (c1070de).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #474   +/-   ##
=======================================
  Coverage   98.54%   98.54%           
=======================================
  Files          19       19           
  Lines        3360     3362    +2     
  Branches      492      493    +1     
=======================================
+ Hits         3311     3313    +2     
  Misses         26       26           
  Partials       23       23           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[raphaelvallat]

Hi @AlexanderJCS — apologies for the long delay. One request to update the tests to make sure this line is covered by the tests. Thanks

[raphaelvallat]

Can you please add a test in https://github.com/raphaelvallat/pingouin/blob/main/tests/test_correlation.py test_corr function:

stats = corr(x, x, method="percbend")  # calls _correl_pvalue
assert np.isclose(stats.at["percbend", "r"], 1)
assert np.isclose(stats.at["percbend", "p_val"], 0)

[AlexanderJCS]

Hi Raphael, I added these lines in the test_corr(self) function under the perfect correlation test for the pearson method.

[AlexanderJCS]

@raphaelvallat The edge case is covered by the correlation test. Please let me know if there's anything else you would like me to change.

[AlexanderJCS]

Hi @raphaelvallat, any updates on the status of this PR?

[raphaelvallat]

Approved and merging now, thanks!