How is the number of subsets and their threshold chosen in a ID3 decision tree classifier?

Question

I am trying to implement an ID3 decision tree classifier. I understand that entropy is used to decide which attribute to base the split on. What I don\'t understand is how t