I would rank my features with the entropy score and select m relevant features. My dataset is unlabeled. How can I use this with python?