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

Skip to content

FIX FeatureUnion for polars output with intermediate duplicated column names #32106

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom
Open
wants to merge 7 commits into
base: main
Choose a base branch
from
Open

FIX FeatureUnion for polars output with intermediate duplicated column names #32106

wants to merge 7 commits into from
+150 −137

Conversation

jeremiedbb
Copy link
Member

@jeremiedbb jeremiedbb commented Sep 4, 2025 •
edited
Loading

Fixes #32104

@glemaitre or @thomasjpfan maybe ?

@github-actions GitHub Actions
Copy link

github-actions bot commented Sep 4, 2025 •
edited
Loading

✔️ Linting Passed

All linting checks passed. Your pull request is in excellent shape! ☀️

Generated for commit: fe0ecb0. Link to the linter CI: here

jeromedockes
jeromedockes reviewed Sep 4, 2025
View reviewed changes
@jeremiedbb @jeromedockes
Update sklearn/pipeline.py
b9e996f 
Co-authored-by: Jérôme Dockès <[email protected]>
@jeremiedbb
Copy link
Member Author

jeremiedbb commented Sep 5, 2025 •
edited
Loading

Hum, actually this fix is not enough. When verbose_feature_names_out=False, it doesn't use get_feature_names_out but keep the current column names instead (i.e. with the meaningless intermediate names that I added here).

So a proper fix would be to do something like in the ColumnTransformer, that is if verbose_feature_names_out=False, obtain each transformer out feature names and use those before stacking, and raise an informative error in case of duplicate column names.

EDIT: Actually that was a bug that I fixed in this PR, see #32106 (comment), so we can keep the simple fix

@jeremiedbb jeremiedbb marked this pull request as draft September 5, 2025 13:07
jeremiedbb
jeremiedbb commented Sep 16, 2025
View reviewed changes
sklearn/utils/_set_output.py
Comment on lines +30 to +34
except AttributeError as e:
if "does not provide get_feature_names_out" in str(e):
return None
else:
raise
Copy link
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Side note for reviewers: this was bypassing the check for duplicates in _add_prefix_for_feature_names_out (itself called in get_feature_names_out), explaining why the error did not happen with pandas even when verbose_feature_names_out was set to False.

@jeremiedbb jeremiedbb marked this pull request as ready for review September 16, 2025 14:51
@jeremiedbb jeremiedbb mentioned this pull request Sep 16, 2025
Open
@jeremiedbb
Copy link
Member Author

I took the opportunity to clean-up the whole logic in the ColumnTransformer as well. There's no need to compute the feature names and do the duplicate columns here. It should be done by the set_output wrapper.

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

@jeromedockes jeromedockes jeromedockes left review comments

Reviewers whose approvals may not affect merge requirements

At least 1 approving review is required to merge this pull request.

Assignees
No one assigned
Labels
module:pipeline
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.

FeatureUnion with polars output can error due to duplicate column names
2 participants
@jeremiedbb @jeromedockes