I know that f_classif method is mostly used for scoring numerical features and categorical targets. But I'm wondering if it's possible to score features coded by integers or binary codes against categorical targets - as the data below:
Feature = [0, 1, 0, 2, 1, 2, ...]
Target = [1, 0, 0, 1, 1, 1, ...]
Also, I have another question - if data includes mix of both categorical and numerical features, what's the best method to score their importance against categorical targets?