You can launch the Jupyter notebook by doing:

jupyter notebook src/bosch.ipynb

Using the Intel Python distribution for a possible performance (training speed) boost:

source activate idp

if the csv files are too long, you can use the split command like this to make them into smaller files:

# in the resources folder
sh make_smaller_files.sh train_numeric.csv

This splits the csv file train_numeric.csv into smaller files with at most 50000 entries each, with the original header replicated in each file. They are named as split_train_numeric_a.csv, split_train_numeric_b.csv, and so on.