I am familiar with basic techniques for binary and multi-class classifications (logistic, LDA, RF, XGboost). But are there techniques for handling classifications where each observation can belong to multiple classes?

Currently, I'm thinking that one possibility would be to transpose the dataset such that the multiple classes become separate single-label observations, and then go with any classification model. Another possibility would be to create multiple binary responses representing whether or not the observation belongs to that class. Then train separate binary classifiers and somehow ensemble the results?

Would either of this work? What else am I missing?

  • $\begingroup$ Look into multi-label classifiers $\endgroup$ – HEITZ Aug 30 '17 at 3:57

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.