I with my BIG team (23 people) ranked 108th (TOP 3%) in the Santander Value Prediction Challenge. It was a fun adventure due to the size of the team, but I made a significant contribution to our result and I want to share some of our code.
BoostARoota and lgb.ipynb
- notebook in which I used a BoostARoota library to select useful features and learning my best LGBM model, which was used in the final ensemblecross validation 20x5.ipynb
- code of our cross validation scheme with 100 foldsfive folds CV (only for fast tests).ipynb
- notebook with simple kfold validation for quick testing of ideasprepare data.ipynb
- notebook with my data processing code (statistical features, space reduction features(PCA, TruncatedSVD, FastICA, FactorAnalysis, SparseRandomProjection, GaussianRandomProjection) and other)