Unit tests for different subsamples #468
Conversation
There was a problem hiding this comment.
Thank you for this PR! Great! @BaptisteCalot
Note that the outputs should be of the same style in MAPIE. So the correction should not be applied here but rather directly in the original file. Note that you should make sure that if they are changes, we should consider why they occur and if they can be resolved without forcing the typing.
mapie/tests/test_subsample.py
Outdated
| length=length, random_state=0) | ||
| trains = [x[0] for x in cv.split(X)] | ||
| tests = [x[1] for x in cv.split(X)] | ||
| for i in range(n_resamplings): |
There was a problem hiding this comment.
Don't forget to use combinations instead of the loop! :)
mapie/tests/test_subsample.py
Outdated
| ) | ||
| trains2 = [x[0] for x in list(cv2.split(X))] | ||
| tests2 = [x[1] for x in list(cv2.split(X))] | ||
| np.testing.assert_array_equal(trains1, trains2) |
There was a problem hiding this comment.
I'd rather have the same test i.e. assert_equal also here
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #468 +/- ##
==========================================
Coverage 100.00% 100.00%
==========================================
Files 39 41 +2
Lines 4616 5106 +490
Branches 487 871 +384
==========================================
+ Hits 4616 5106 +490 ☔ View full report in Codecov by Sentry. |
Description
This PR aims to create unit tests to verify that the train and test sets of the split method from Subsample and BlockBootstrap classes are different even though the random state is the same
Fix issue #290
Type of change
How Has This Been Tested?
Checklist : OK
make lintmake type-checkmake testsmake coveragemake doc