tdhd/interpretability-class

interpretability-class

This repo contains exercises regarding interpretability of ML models. There are some older ones in exercises, the newer more tooling focused notebook and setup guide can be found in tooling-exercises folder.

setup

To get started, create a new conda environment and install all of the dependencies, run

conda create --name interpret python=3.7
conda activate interpret
conda install tensorflow matplotlib scikit-learn seaborn numpy pandas scipy statsmodels ipython jupyter
pip install eli5 lime keras-applications

For some functionality you might need to install graphviz.