This folder contains our code related to feature engineering and training for the final 5 Gradient Boosted (GBT) models. The majority decision of these models give our final skip prediction.
By running 'run_all.sh' script you can reproduce our results.
bash run_all.sh DATA_DIR EXPERIMENT_DIRYou have to specify the following 3 parameters:
- DATA_DIR: the root folder where your data and pre-generated autoencoder and skip pattern based features reside
- EXPERIMENT_DIR: the folder where you want to save your models, and predictions
- MAX_THREADS: the maximum number of threads you can allow to run this code (specify this in the 'run_all.sh' script)