MNT remove default behaviour deprecation from class_likelihood_ratios #31331
Conversation
| @@ -2220,28 +2213,11 @@ class are present in `y_true`): both likelihood ratios are undefined. | |||
| "`UndefinedMetricWarning` will always be raised in case of a division by zero " | |||
| "and the value set with the `replace_undefined_by` param will be returned." | |||
There was a problem hiding this comment.
We also talked about expanding this error message to include a hint on how to handle this with catch_warnings
There was a problem hiding this comment.
Yes, I've expanded the UndefinedMetricWarning messages to mention warnings.catch_warnings().
|
There's probably also a changelog for the old PR which we need to fix? |
|
Thanks, @adrinjalali. |
I think that #29288 is missing a changelog entry for the deprecation of |
There was a problem hiding this comment.
(I added the missing changelog entry)
LGTM. Thanks @StefanieSenger
|
Thank you, @jeremiedbb! |
Reference Issues/PRs
towards #29048
(It's a partial reversal of #29288.)
What does this implement/fix? Explain your changes.
This removes the deprecation of the default behaviour in case of a zero division from
class_likelihood_ratios.The default behaviour in case of a division by zero was
np.nanbefore, as (indirectly) defined byraise_warning=True, which was the default before..CC @adrinjalali, @jeremiedbb, @virchan