0

I have read this to learn about various method in multi-label classifiers. I learned that there are 3 techniques to do multi-label classifications:

1.Problem Transformation
2.Adapted Algorithm
3.Ensemble approaches

In the category of Problem transformations there are more three sub categories:

a.Binary Relevance
b.Classifier Chains
c.Label Powerset

I know that when we want better result we should apply the ensemble model. I would like to know in which situations the other different algorithm we should use.

I know how they differently work, but I do not know when I should use each of them.

And Also there is only two method implemented for Adapted Algorithm. what if I want other methods but implemented in adapted algorithm approach?

Please let me know if my statements are not clear.

Thanks,

sariii
  • 2,020
  • 6
  • 29
  • 57
  • 1
    I am not quite sure what you are asking. What do you mean by "consider the whole labels with each other"? – Vivek Kumar Sep 04 '18 at 07:15
  • @VivekKumar thank you so much for following with me. I mean for multi-label classification which approach makes more sense in terms of giving the correct result. In my idea the case in which consider the binary is not realistic. – sariii Sep 05 '18 at 05:22
  • Or can you please summarize with these different approach, in which application which one should be applied? – sariii Sep 05 '18 at 05:24
  • @VivekKumar Kindly I have updated my question. and I think it now makes more sense :) – sariii Sep 05 '18 at 05:32

0 Answers0