119th solution for Porto Seguro's Safe Driver Prediction on Kaggle (https://www.kaggle.com/c/porto-seguro-safe-driver-prediction).
Final submission is the rank averaging of the following 4 models (2 original models and 2 kernel models).
- l1_lgb (Public LB: 0.285)
- l1_xgb (Public LB: 0.285)
- Base on Froza & Pascal single XGB LB (0.284)
- RGF + Target Encoding + Upsampling
To train the original models and predict, run:
python l1_lgb.py
python l1_xgb.py
To average the predictions, run:
python l2_rank_avg.py