Should I balance class weights on logistic regression?
I check pros and cons about using class_weight='balanced' in scikit learn's logistic regression. I share resources/papers written by others and I do a minimal testing with synthetic data. I also plot a decision-curve-analysis of the resulting models.