Multi-Class or Multi-Label Classification with LightGBM

Question

I am working on a classification project that an outcome may belong to multiple classes. For example, the outcomes may belong to Class A, B, and/or C; e.g., A, B, A&B, A&C, B&C, etc. However, I want to predict the probability of a class. For example, P(A)=Prob of outcome contains Class A; e.g., Pr(A)+Pr(A&B)+Pr(A&C)+Pr(A&B&C).

I prefer using LightGBM for it. My questions are:

1. For LightGBM, do outcome classes have to be mutually exclusive?
2. For LightGBM, what is the difference between multiclass vs multiclassova?
3. If I model it as a multi-label classification (e.g., N-class) problem, is it equivalent to 2^N multi-class classification and each class is one-hot encoded of N classes?

Answer 1

4 years later: I have run into a similar problem where I want to train a multi-label classification model using lightgbm/ xgboost. To the best of my knowledge, there still isn't one for lightgbm but you can implement it on xgboost: https://xgboost.readthedocs.io/en/stable/tutorials/multioutput.html

So the answer to your question 1 is yes, for lightgbm the labels have to be mutually exclusive.