Description for the competition in russian: https://mlbootcamp.ru/round/12/rating/

This version is for 62th place (final logloss: 0.5313173) that I created on 21.06.2017 (the competition went on from 15.06.2017 to 15.07.2017).

This is a Jupyter Notebook with a machine learning model using XGBoost that is trying to predict a diagnosis for a cardiovascular disease depending on a patient parameters like height, weight, arterial pressure, cholesterol and glucose level, sports activity etc.

It doesn't go really well as you can see with the logloss score, so I do not recommend using it on a real case, especially when the data is that "dirty" (who the hell are those people with 16020/70 arterial pressure?!)