Recourse-FARE: (Explainable) Algorithmic Recourse with Reinforcement Learning and MCTS

This library provides a set of methods which can be used to achieve model-agnostic algorithmic recourse given a black-box model. The library enables customization of all the aspects of the recourse process, from the actions available to the models employed.

If you want to have a gist of a practical application and how to use recourse-fare, please have a look at the following notebook Tutorial: training FARE and E-FARE models on the Adult dataset

Install

The library can be easily installed from GitHub directly. We suggest using a virtualenv to make it easier to develop on top of it.

pip install git+https://github.com/unitn-sml/recourse-fare.git@v0.1.0

Development

We can easily download the following library and install it locally in your preferred (virtual) environment. We suggest using Python 3.7 and conda. If you find any issue with the following procedure, feel free to open an issue!

git clone https://github.com/unitn-sml/recourse-fare.git
cd recourse-fare
conda create --name recourse_fare python=3.7
conda activate recourse_fare
pip install -e .

References

We use the library in the following projects:

[1] De Toni, Giovanni, Bruno Lepri, and Andrea Passerini. "Synthesizing explainable counterfactual policies for algorithmic recourse with program synthesis." Machine Learning (2023): 1-21, 10.1007/s10994-022-06293-7

[2] De Toni, Giovanni, et al. "User-Aware Algorithmic Recourse with Preference Elicitation." arXiv preprint arXiv:2205.13743 (2022), 2205.13743