Decision Tree Classifier with Majority vote (Is 50% a majority vote)

Question

I have a homework question that requires me to create a 3 level tree (root, intermediate and leaf) with majority vote and expand the tree if 3 level tree is not possible. I have to use ID3 algorithm. Its a manual question rather than coding assignment and dataset contains 16, 4 features and a class label with T, F as their values. I started calculating entropies and information gain for different nodes. When I complete the 3 level tree, I can assign the classes at the leaf with 50%, 100%, 100% and 75% majority vote or confidence. My questions are

1. is 50% really a majority vote?
2. Do I need to expand on that part of the tree that has 50% confidence
3. or I can just assign the classes arbitrarily?

Regards

Answer 1

The goal of ID3 is to get the purest nodes possible ( ironically that is what contributes to its problem of overfitting), so 50% is not pure at all, the data under that node is equally likely to be in one of the classes which makes peedicition tricky, it would be better to grow the tree further and find nodes which are more pure than atleast 50%.