MNT Refactor scorer using _get_response #18589
Conversation
| f"pos_label={pos_label} is not a valid label: {classes}" | ||
| ) | ||
|
|
||
| def _select_proba_binary(self, y_pred, classes): |
There was a problem hiding this comment.
This function was moved within _get_response when selecting the binary probability.
cc27a27 to
a212a81
Compare
|
@ogrisel @thomasjpfan @NicolasHug I think this is ready for a review. |
|
I added a summary of what was done in the welcoming message of this PR. |
|
OK I think this is good for a new review @NicolasHug @ogrisel |
| """Call estimator with method and args and kwargs.""" | ||
| response_method = kwargs["response_method"] |
There was a problem hiding this comment.
Can update the scorers in a follow up PR so this PR can be a little shorter?
sklearn/utils/validation.py
Outdated
| Parameters | ||
| ---------- | ||
| estimator : estimator instance | ||
| Classifier to check. |
There was a problem hiding this comment.
This is used for regressors and classifiers.
|
There's been no activity here since February when you added this to 1.0 @glemaitre . Do you think this can be merged in time for the release? |
|
I think it is best to have all plotting display in and then do the refactoring. The refactoring is not a blocker, this is just maintenance.
Sent from my phone - sorry to be brief and potential misspell.
|
|
this pr is superseded. I'm removing it from the milestone and adding the more recent one instead |
|
Since this PR is superseded, can we close it? It would be easier to figure out which one is the "current one". |
|
Superseded by #20999. |
Supersede #18212
Refactor the scorer such that they make use of
_get_responsealready define in the plotting function.This refactoring can also be beneficial for #16525.
Summary of what was done:
_check_response_method. Its job is to return the method of a classifier or a regressor to later predict. If the method does not exist, it raises an error. This function was already existing indeed._get_response. A function that returns the prediction depending on the response method. We take into account thepos_label. Thus, it will allow to not make any mistake in the future by forgetting to inverse the decision function or select the right column of probabilities for binary classification. We hard-coded this behaviour in a lot of different places and this function reduces the amount of redundant code.