search

trevismd / statannotations

add statistical significance annotations on seaborn plots. Further development of statannot, with bugfixes, new features, and a different API.
Other
624 stars 67 forks source link

better implementation to hide non significant results #95

Closed sepro closed 1 year ago

sepro commented 1 year ago

Following the discussion with @trevismd here the parameter show_non_significant has been refactored to hide_non_significant, threshold is now correctly based on alpha and (corrected) p-value. Type of the result instance is checked and a warning is thrown when this is not a StatResult.

I've tested this locally and it works, though a unit test still needs to be added cover these few lines of code. How would you like this to be tackled.

codecov[bot] commented 1 year ago

Codecov Report

Merging #95 (7eebf7c) into dev (2a5ebe1) will decrease coverage by 0.00%. The diff coverage is 96.55%.

@@            Coverage Diff             @@
##              dev      #95      +/-   ##
==========================================
- Coverage   97.04%   97.04%   -0.01%     
==========================================
  Files          29       29              
  Lines        1898     1927      +29     
==========================================
+ Hits         1842     1870      +28     
- Misses         56       57       +1
Impacted Files Coverage Δ
statannotations/Annotator.py 91.48% <66.66%> (-0.18%) :arrow_down:
statannotations/Annotation.py 100.00% <100.00%> (ø)
statannotations/stats/StatResult.py 92.30% <100.00%> (+0.81%) :arrow_up:
tests/test_annotation.py 100.00% <100.00%> (ø)
tests/test_stat_result.py 100.00% <100.00%> (ø)

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

sepro commented 1 year ago

Refactored code to make it test-able and added tests to have full coverage on the new lines. @trevismd could you check the logic in stats.StatResult.is_significant() ? I'm not 100% sure if self._corrected_significance is used as intended here.

sepro commented 1 year ago

@trevismd finished refactoring the code in a few spots to get it working again after pulling in your suggestions. Also split the unit-test and put functions in alphabetical order per your other suggestions.

Think that covers all comments.

Happy holidays !

M-Hisham commented 1 year ago

I tested it, it works like a charm

Thanks Sepro for the nice feature