Explainable Machine Learning Challenge

This is my proposal for the explainable.ml challenge organised by FICO, Google, NIPS.

Here is the paper which details my approach.

The "Global interpretation - intuitive.ipynb" notebook in this repository contain an interactive visualization fit for fast intuitive data exploration.

The "RuleFitCustom - Global and Local explanations.ipynb" notebook contains local and global rules that are simpler than extracted from standard RuleFit.

Installation instructions

We recommend Python 3.7 and an environment with the following packages:

pandas matplotlib plotly ipython ipykernel

Installation of the customized implementation of RuleFit:

pip install -e ./rulefitcustom

---

For consulting on explainable machine learning, please contact me at at http://benoit.paris