Measure the ability of variables to differentiate between two groups

Question

I have binary, categorical, interval, and metric variables. For each of them, I would like to find a measure that helps me decide on how well the variable is able to help predicting a binary target variable, i.e. whether considering the variable adds value or not.

I thought of using entropy/information gain or mere relative frequencies. For metric data, I would need to bin values, however.

What kind of literature would help me find appropriate measures? What would you suggest?

Answer 1

This is too broad to be answered here.

In essence, you are asking about feature selection, a whole research subdomain: which features (attributes) to use, and in which priority.