[MRG+1] check that splitters handle 2d y and give reasonable errors on multilabel y #9744
Conversation
| y_2d = y.reshape(-1, 1) | ||
| y_multilabel = rng.randint(0, 2, size=(n_samples, 3)) | ||
| groups = rng.randint(0, 3, size=(n_samples,)) | ||
| splitters = [LeaveOneOut(), LeavePOut(p=2), KFold(), StratifiedKFold(), |
There was a problem hiding this comment.
should this be a global or generated by a global function?
There was a problem hiding this comment.
I was considering it. A global variable seems consistent with other lists of classes, but I didn't do that given that it's only used once so far.
| y_multiclass_2d = y_multiclass.reshape(-1, 1) | ||
| cv = check_cv(3, y_multiclass_2d, classifier=True) | ||
| np.testing.assert_equal(list(StratifiedKFold(3).split(X, y_multiclass_2d)), | ||
| list(cv.split(X, y_multiclass_2d))) |
There was a problem hiding this comment.
it's worth testing that it differs from KFold(3) to be sure that the assertion is meaningful
There was a problem hiding this comment.
assertion is copied from above. Should I add a check there, too?
|
LGTM |
|
LGTM, merging, thanks a lot! |
|
Now that we've merged this, I'm not sure why we should be stopping users have multilabel y in splitters that do not depend on y's data, such as |
|
Sorry, my mistake. This is fine. |
Fixes #9742.