Thanks to visit codestin.com
Credit goes to github.com

Skip to content

TST Remove non-functional test_ridge_loo_cv_asym_scoring (#16041)#34422

Draft
dhruv7477 wants to merge 2 commits into
scikit-learn:mainfrom
dhruv7477:fix/16041-ridge-loo-asym-scoring-test
Draft

TST Remove non-functional test_ridge_loo_cv_asym_scoring (#16041)#34422
dhruv7477 wants to merge 2 commits into
scikit-learn:mainfrom
dhruv7477:fix/16041-ridge-loo-asym-scoring-test

Conversation

@dhruv7477

@dhruv7477 dhruv7477 commented Jul 2, 2026

Copy link
Copy Markdown
Contributor

Reference Issues/PRs

Fixes #16041.

What does this implement/fix?

Removes test_ridge_loo_cv_asym_scoring. As reported in #16041, it did not test anything meaningful: it compared RidgeCV(cv=n_samples, scoring="explained_variance") against the GCV path, but with leave-one-out each fold has a single sample and explained_variance on one sample is degenerate, so every alpha tied and the test passed by chance. Per maintainer suggestion, removing it rather than rewriting.

@dhruv7477

Copy link
Copy Markdown
Contributor Author

This is a test-only fix (correcting test_ridge_loo_cv_asym_scoring) with no user-facing behavior or API change, so no changelog entry is needed. Could a maintainer add the No Changelog Needed label to satisfy the changelog check?

@ogrisel

ogrisel commented Jul 3, 2026

Copy link
Copy Markdown
Member

I think we should just remove this useless test.

@dhruv7477 dhruv7477 force-pushed the fix/16041-ridge-loo-asym-scoring-test branch from eacff42 to 8726ea7 Compare July 3, 2026 12:28
@dhruv7477

Copy link
Copy Markdown
Contributor Author

@ogrisel Agreed — removed the test. Thanks!

@dhruv7477 dhruv7477 changed the title TST Fix RidgeCV leave-one-out asymmetric-scoring test (#16041) TST Remove non-functional test_ridge_loo_cv_asym_scoring (#16041) Jul 3, 2026
@dhruv7477

Copy link
Copy Markdown
Contributor Author

@ogrisel Done — removed the test as suggested; it's now a pure deletion. I've left it as a draft because I've hit the "ready for review" PR limit (my other open PR #33879 is using the slot), so I can't flip this to ready myself. Since this is just the removal you requested, could a maintainer mark it ready / take it from here? Happy to adjust anything.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Wrongly implemented test in RidgeCV

3 participants