run code from main.py for training, this uses ./dataset/train.csv and ./dataset/test_with_solutions.csv test data must be in ./dataset/test.csv results will be in rez.csv code needs python 2.7, scikit-learn 0.12 (along with numpy, scipy, ..), neurolab, nltk, maybe others :) 4gb of ram are needed also, a fast cpu is recommended - ~30 minutes runtime on core i7