I found this question but I need an answer to the other direction.

Example: Let's say we want to predict if a person with a certain profile wants to buy product A and/or B. So we have 2 binary classes A and B that don't exclude each other:

A     B
0     1
1     1
0     0
1     0

(We don't want to predict how likely it is for a person to buy B if the person has already bought A.) Does it - in general or under certain conditions - make sense to transform this problem into a single-class multi-label problem with 4 labels (4=#combinations of A and B)? What if the number of binary classes is larger than 2?


It depends on the contextual link between A and B.

If they are completely different categories with no or low correlation, there shouldn't be necessary to have a single class multi label.

But if A and B are somehow connected, overall if they can represent a scale together (i.e. AB = [0 0] = 0 = "low impact" or AB = [1 1] = 3 = "high impact"), it could be meaningful to have a single class multi label. It all depends on the correlation and the business point of view.

If you can give more information about the purpose of A and B, I might be able to give more information.

  • $\begingroup$ Thanks for your answer. I edited my question to give more information but as far as I can conclude from your answer, in my case it doesn't make sense to transform the problem to a multi-label problem. $\endgroup$
    – LineBreak
    Jan 12 at 18:09
  • $\begingroup$ Yes indeed. In addition to that, multilabel single class shall sooner or later be transformed in numeric values in order to be understood by algorithms (or I don't know another method). As A and B are independent, it doesn't make sense to have a ranged value, otherwise it would give the data a wrong meaning. Consequently, you should keep A and B as binary independent values. $\endgroup$ Jan 13 at 8:23

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.