Mars Spectrometry 2: Gas Chromatography challenge on DrivenData

This presents the implementation of my solution to the challenge (18th place).

Requirements

The packages necessary to run the code are listed below.

pip install -r requirements.txt

Data structure

The project is organized as follows:

gcms_data
+-- raw data
+-- features
+-- models
+-- predictions

Generate submissions

There are two ways to run the code for this project:

• sandalone notebook gcms.ipynb, or
• .py scripts in order to do the pre-processing, hyperparameters tuning, training, and predictions tasks:

python preprocessing.py
python hyperparameter_tuning.py
python training.py
python ensemble_and_submit.py