Baseline - simple.ipynb contains a naive xgboost solution and creates a model file for Track 2
Baseline - advanced.ipynb contains a primer on jagged arrays processing and creates a model file for Track 2
track_2_baseline_advanced_cpp contains an example submission for
Track 2 with catboost and jagged arrays in C++. Run make all to
build and the baseline binary will be accepting data as .csv (not
gzipped) at the standard input and writing predictions to the standard
output. Run make compute_features to build an utility binary that
preprocesses data and outputs the computed features.
track_2_baseline_simple_python contains an example submission for
Track 2 with xgboost and python. Quite slow to read the data and
doesn't have the jagged arrays.
scoring.py contains an implementation of the quality metric
utils.py contains data reading and jagged arrays processing utilities
Evaluate advanced baseline.ipynb compares Python and C++ predictions for Track 2