MAINT Parameters validation for metrics.top_k_accuracy_score #25828
Conversation
sklearn/metrics/_ranking.py
Outdated
| { | ||
| "y_true": ["array-like"], | ||
| "y_score": ["array-like"], | ||
| "k": [Integral], |
There was a problem hiding this comment.
why not limit usage to >1
i.e. "k": [Interval(Integral, 1, None, closed="left"), None],
There was a problem hiding this comment.
Indeed, k cannot be negative. We should use an interval here. However, we cannot use None because we expect only integral values.
There was a problem hiding this comment.
None is fine since closed="left"
There was a problem hiding this comment.
Ups, I misread. I read that it was an option None (not inside the interval). My bad.
There was a problem hiding this comment.
Ah no you're right. I did misread and didn't see the additional None outside the interval :)
There was a problem hiding this comment.
So I misread your misreading :)
There was a problem hiding this comment.
Thank you for all the suggestions and help!
…learn#25828) Co-authored-by: wishyut.pitawanik <wpitawanik@mpiwg-berlin.mpg.de> Co-authored-by: Jérémie du Boisberranger <34657725+jeremiedbb@users.noreply.github.com>
Towards #24862
metrics.top_k_accuracy_scoremetrics.top_k_accuracy_scoreintest_public_functions.pyNote: the
kparameter also works for integersk <= 0, which may not be its intended use case.