TST Add test for documentation of the classes_ attribute #16277
Conversation
|
Nice first step, as you can see in https://dev.azure.com/scikit-learn/scikit-learn/_build/results?buildId=12718&view=ms.vss-test-web.build-test-results-tab, the failed tests are the ones I was talking about in #16273. |
|
Just to be clear, I would suggest fixing the tests in this PR, for CheckingClassifier, it can be skipped because it is only used for tests, so it is probably not necessary to "force" some documentation on it. |
|
contributes to #14312 for HistGradientBoostingClassifier attribute classes_ |
Co-Authored-By: Alexandre Gramfort <[email protected]>
sklearn/utils/_mocking.py
Outdated
|
|
||
| Attributes | ||
| ---------- | ||
| classes_ : array, shape = (n_classes,) |
There was a problem hiding this comment.
That is the only thing I am not sure about ...
Should we "force" the documentation on some classes that are only used for tests like CheckingClassifier? I would be enclined to skip the test instead but I don't have a strong opinion.
An alternative is to do a minimal documentation as was done for the parameters a few lines above, i.e.:
Attributes
----------
classes_
There was a problem hiding this comment.
ok. Definitely more consistent with the parameter documentation. Doing it right now.
don't know about skipping the test.
There was a problem hiding this comment.
The minimal documentation approach is simpler, so let's do this and see whether others have a different opinion.
…ithub.com/judithabk6/scikit-learn into add_test_doc_classes_attribute_estimators
…n#16277) Co-authored-by: Alexandre Gramfort <[email protected]>
…n#16277) Co-authored-by: Alexandre Gramfort <[email protected]>
adding a test to ensure the classes_ attribute is documented in all classifiers.
towards #16273
was originally implemented in PR #12509 by @jrbourbeau