[MRG + 3] OneVsRestClassifier: don't expose predict_proba and decision_function if base estimator doesn't support them #7812
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
methods if they are not supported by base estimator.
amueller
reviewed
Nov 2, 2016
| decision_only = OneVsRestClassifier(svm.SVR()).fit(X_train, Y_train) | ||
| assert_raises(AttributeError, decision_only.predict_proba, X_test) | ||
| assert not hasattr(decision_only, 'predict_proba') |
There was a problem hiding this comment.
these are pytest style asserts, while we are still using nosetests. Maybe use "assert_true" and "assert_false" until we transition?
There was a problem hiding this comment.
a good point; it seems I got too used to pytest asserts :)
|
I'm +1 for this change and it LGTM. |
|
@jnothman a nice catch; I thought about a fallback when writing this PR, but failed to come up with an example where checking self.estimator doesn't work. I'll fix that. |
|
Does it look better now? |
+1 for 0.18.1 as this is kind of a bug fix right? |
raghavrv
reviewed
Nov 7, 2016
|
|
||
| # Estimator which can get predict_proba enabled after fitting | ||
| gs = GridSearchCV(svm.SVC(probability=False), | ||
| param_grid={'probability': [True]}) |
|
Thanks @kmike |
sergeyf
pushed a commit
to sergeyf/scikit-learn
that referenced
this pull request
Feb 28, 2017
…n_function if base estimator doesn't support them (scikit-learn#7812) * OneVsRestClassifier: don't expose predict_proba and decision_function methods if they are not supported by base estimator. * TST use nose-style asserts * handle a case where classifier get predict_proba method only after .fit
Sundrique
pushed a commit
to Sundrique/scikit-learn
that referenced
this pull request
Jun 14, 2017
…n_function if base estimator doesn't support them (scikit-learn#7812) * OneVsRestClassifier: don't expose predict_proba and decision_function methods if they are not supported by base estimator. * TST use nose-style asserts * handle a case where classifier get predict_proba method only after .fit
NelleV
pushed a commit
to NelleV/scikit-learn
that referenced
this pull request
Aug 11, 2017
…n_function if base estimator doesn't support them (scikit-learn#7812) * OneVsRestClassifier: don't expose predict_proba and decision_function methods if they are not supported by base estimator. * TST use nose-style asserts * handle a case where classifier get predict_proba method only after .fit
paulha
pushed a commit
to paulha/scikit-learn
that referenced
this pull request
Aug 19, 2017
…n_function if base estimator doesn't support them (scikit-learn#7812) * OneVsRestClassifier: don't expose predict_proba and decision_function methods if they are not supported by base estimator. * TST use nose-style asserts * handle a case where classifier get predict_proba method only after .fit
maskani-moh
pushed a commit
to maskani-moh/scikit-learn
that referenced
this pull request
Nov 15, 2017
…n_function if base estimator doesn't support them (scikit-learn#7812) * OneVsRestClassifier: don't expose predict_proba and decision_function methods if they are not supported by base estimator. * TST use nose-style asserts * handle a case where classifier get predict_proba method only after .fit
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Currently OneVsRestClassifier has predict_proba and decision_function methods even if base estimator doesn't have them. This makes it harder to check if an estimator supports predict_proba or not because it is not enough to run
hasattr(estimator, 'predict_proba'), one have to special-case OneVsRestClassifier.In this PR it is changed, so that OneVsRestClassifier is more similar to Pipeline: if base estimator doesn't provide predit_proba or decision_function then OneVsRestClassifier also doesn't provide them.
It is not 100% backwards compatible: AttributeError is raised earlier, on ovr.decision_function / ovr.predict_proba access, not when these methods are called.