1. Valid the model run the 'run_test.py' 2. Train another model run the 'run_train.py' 3. Other requirements (1) data feature put in the 'data/feature' directory, including 5 csv files, which are pretrained offline (2) environment python == 3.6 pip install catboost xgboost scipy sklearn pandas 4. Checkpoints and result The submission file is put in 'results' directory The model file is put in 'checkpoint' directory