This project contains my proceedings dealing with the porto-seguro safe driver competition launched on Kaggle, which ended Nov 2017. My score is not a big deal, though I'd like to highlight in this repo the way one can write python classes for feature engineering.
Add input and out folder
>>> mkdir input
>>> mkdir out
Make sure, that you have the correct train and test data downloaded from kaggle's competition site and saved in the input folder.
Starting the project with
>>> python main.py
This will load the dataset from input folder, does a bit feature engineering and trains a LGBM model whose output will be saved to the out folder.