[MRG] Fix sklearn.metrics.classification._check_targets where y_true and y_pred are both binary but the union is multiclass #8377
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
lesteve
commented
Feb 16, 2017
| if y_type in ["binary", "multiclass"]: | ||
| y_true = column_or_1d(y_true) | ||
| y_pred = column_or_1d(y_pred) | ||
| if y_type == "binary": |
Member
Author
There was a problem hiding this comment.
This is the less invasive fix I could find. This is not super efficient because np.unique has been already called on y_true and y_pred to figure out their type in type_of_target.
294e698 to
e01f779
Compare
lesteve
commented
Feb 16, 2017
| y_true_inv = ["b" if i == "a" else "a" for i in y_true] | ||
|
|
||
| assert_almost_equal(matthews_corrcoef(y_true, y_true_inv), -1) | ||
| y_true_inv2 = label_binarize(y_true, ["a", "b"]) * -1 |
Member
Author
There was a problem hiding this comment.
y_true_inv2 is an array of 0 and 1 whereas y_true is an array of 'a' and 'b'. The test was passing just by chance as noted in #8094 (comment).
lesteve
commented
Feb 16, 2017
| # And also for any other vector with 0 variance | ||
| mcc = assert_warns_message(RuntimeWarning, 'invalid value encountered', | ||
| matthews_corrcoef, y_true, | ||
| rng.randint(-100, 100) * np.ones(20, dtype=int)) |
Member
Author
There was a problem hiding this comment.
Similar to the previous change, y_true is an array of 'a' and 'b' and we are trying to compare it to an array of ints, so something somewhere should tell us that something is wrong.
Codecov Report
@@ Coverage Diff @@
## master #8377 +/- ##
==========================================
+ Coverage 94.75% 94.75% +<.01%
==========================================
Files 342 342
Lines 60902 60911 +9
==========================================
+ Hits 57708 57717 +9
Misses 3194 3194
Continue to review full report at Codecov.
|
jnothman
reviewed
Feb 21, 2017
Member
jnothman
left a comment
There was a problem hiding this comment.
LGTM. A bug fix changelog entry is deserved.
but the union of them is multiclass.
37474b1 to
b71c3b4
Compare
Member
Author
|
Thanks for the review, I added an entry in the changelog. |
MechCoder
approved these changes
Apr 21, 2017
Member
|
Thanks you. |
Member
Author
|
Thanks for the review @MechCoder ! |
Sundrique
pushed a commit
to Sundrique/scikit-learn
that referenced
this pull request
Jun 14, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
dmohns
pushed a commit
to dmohns/scikit-learn
that referenced
this pull request
Aug 7, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
dmohns
pushed a commit
to dmohns/scikit-learn
that referenced
this pull request
Aug 7, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
NelleV
pushed a commit
to NelleV/scikit-learn
that referenced
this pull request
Aug 11, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
paulha
pushed a commit
to paulha/scikit-learn
that referenced
this pull request
Aug 19, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
maskani-moh
pushed a commit
to maskani-moh/scikit-learn
that referenced
this pull request
Nov 15, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
jwjohnson314
pushed a commit
to jwjohnson314/scikit-learn
that referenced
this pull request
Dec 18, 2017
…and y_pred are both binary but the union is multiclass (scikit-learn#8377) * Fix _check_targets where y_true and y_pred are both binary but the union of them is multiclass. * Add entry in changelog
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
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.
Reference Issue
Fix part of #8098.
What does this implement/fix? Explain your changes.
In the case where both
y_predandy_trueare binary it checks that the union of them only has two values otherwise returns'multiclass'as type.Any other comments?
I saw #8094 and thought that it was too many different things as once. This is an attempt of making it do one less thing.