Cross-validation broken for multilabel classifiers. #3405

Closed
adarob opened this Issue Jul 16, 2014 · 2 comments

Comments

Projects
None yet
2 participants

adarob commented Jul 16, 2014

label_test_folds = test_folds[y == label]

The above line in cross_validation.py is not compatible with arrays of tuples or the 2D arrays produce by MultiLabelBinarizer.

Owner

arjoly commented Jul 16, 2014

Could this #3128 be a fix for your use case?

adarob commented Jul 16, 2014

It is indeed. Thanks.

@adarob adarob closed this Jul 16, 2014

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment