my old profile

random sampling: https://stackoverflow.com/questions/36618259/how-to-use-pythons-fastfm-library-factorization-machines-for-recommendation-t

Also I have date in data. So should I add negative examples by date? What strategies are there?