it may be interesting for you to know that there is discussion https://github.com/scikit-learn/scikit-learn/issues/2662 about your approach as I can see some say it is not machine learning I try defend you... https://github.com/jnothman https://github.com/remiadon https://github.com/rmenich https://github.com/aloknayak29 https://github.com/actsasgeek https://github.com/larsmans https://github.com/jamesmcm