1

In scikit-learn, there is a strategy called sklearn.multiclass.OneVsRestClassifier, which can be used for both multiclass and multilabel problems. According to its documentation:

"In the multilabel learning literature, OvR is also known as the binary relevance method".

My question is,

Is there is any difference between this scikit-learn strategy and skmultilearn.problem_transform.BinaryRelevance?

Thank you in advance.

1 Answers1

1

No, there is no difference. They work in the exact same way. They both predict probabilities for an instance belonging to a class, yes or no.

Joshua1990
  • 26
  • 2