skmultilearn.base.problem_transformation module

class skmultilearn.base.problem_transformation.ProblemTransformationBase(classifier=None, require_dense=None)[source]

Bases: skmultilearn.base.base.MLClassifierBase

Base class providing common functions for multi-label classifiers that follow the problem transformation approach.

Problem transformation is the approach in which the original multi-label classification problem is transformed into one or more single-label problems, which are then solved by single-class or multi-class classifiers.

Scikit-multilearn provides a number of such methods:

  • BinaryRelevance - performs a single-label single-class classification for each label and sums the results BinaryRelevance
  • ClassifierChains - performs a single-label single-class classification for each label and sums the results ClassifierChain
  • LabelPowerset - performs a single-label single-class classification for each label and sums the results LabelPowerset
Parameters:
  • classifier (scikit classifier type) – The base classifier that will be used in a class, will be automagically put under self.classifier for future access.
  • require_dense (boolean (default is False)) – Whether the base classifier requires input as dense arrays.