✔️ Linting Passed

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

_{Generated for commit:89a8cfc. Link to the linter CI:here}

Thanks a lot@natmokval! 🚀

I think we can now fully remove the definedmax_error_scorer in the_SCORERS dict. Apart from that little thing, I think that clean-up is fine and complete.

Thanks@StefanieSenger for reviewing this PR. I couldn’t findmax_error_scorer in the_SCORERS dictionary — it looks like it was removed during deprecation.

That looks very good. Thanks again,@natmokval! Approving.

Would you be so kind to help me with understanding something? I wonder which AI tool you were using and how it worked to produce this comment:

Thanks@StefanieSenger for reviewing this PR. I couldn’t find max_error_scorer in the _SCORERS dictionary — it looks like it was removed during deprecation.

Since you have just removedmax_error_scorer from the_SCORERS dict with your last commit89a8cf, it obviouslywas there and had not already been removed with the deprecation. Therefore, I find this comment confusing.

When in the process was it produced and by which tool? Thanks for your help.

Thanks@StefanieSenger for the comment. I used a tool called Natural Intelligence.
Here's a part of thediff between main and the commit with the deprecataion

@@ -850,7 +867,7 @@ fowlkes_mallows_scorer = make_scorer(fowlkes_mallows_score) _SCORERS = dict(     explained_variance=explained_variance_scorer,     r2=r2_scorer,-    neg_max_error=neg_max_error_scorer,+    max_error=max_error_scorer,

I understand this as removingmax_error_scorer from the_SCORERS, of course I can be wrong.

Unfortunately, I did not find a way to show code changes from another PR directly in my comment, but you can see that I mean here:
https://github.com/scikit-learn/scikit-learn/pull/29462/files#diff-dfa830b15d3c3d544f99c5d022b7fb6067d97f3ff0289002291e78b65b02ec58L867-L870

I'm very sorry,@natmokval, my bad, I didn't even mean_SCORERS but the defined list of standard regression scores. I had mistakenly thought it was in_SCORERS, probably because I clicked around too much. The generated message (with the long — dash) led me to think it was automated and an llm hallucinating, which is what I am seeing several times a day in PRs that were not checked too eagerly by their authors because they trust their llms too much. I should have checked this better.

Thanks for sharing the tool you used. It helps me understand how these kind of llm-generated messages come into being.

I'm very sorry,@natmokval, my bad, I didn't even mean_SCORERS but the defined list of standard regression scores. I had mistakenly thought it was in_SCORERS, probably because I clicked around too much. The generated message (with the long — dash) led me to think it was automated and an llm hallucinating, which is what I am seeing several times a day in PRs that were not checked too eagerly by their authors because they trust their llms too much. I should have checked this better.

Thanks for sharing the tool you used. It helps me understand how these kind of llm-generated messages come into being.

No problem@StefanieSenger, I understand.

Thanks for the PR@natmokval. We usually tend to do this clean-up PRs closer to the release to prevent having to deal with painful conflicts when doing intermediate bug-fix releases. For instance, 1.8 is expected in November so there's a good chance that we release a 1.7.2 before that.

I'd rather wait until we're close to 1.8, i.e. October or September, to merge this PR. I'm putting it into the 1.8 milestone so that we don't forget it.

We usually tend to do this clean-up PRs closer to the release to prevent having to deal with painful conflicts when doing intermediate bug-fix releases.

I wasn't aware of this either. Thank you,@jeremiedbb.