recourse-game is a Python module designed for algorithmic recourse built on top of Scikit-Learn and is distributed under the MIT license.
This repository contains an implementation of the agent-based simulation framework proposed in "Setting the Right Expectations: Algorithmic Recourse Over Time" by Fonseca et al.
This repo is currently under active development and more features are expected to be added in the future.
recourse-game requires:
- Python (>= )
Recourse-learn plotting capabilities (i.e., the recgame.visualization
submodule) require Matplotlib (>= ). To use the DiCE algorithm, the library
dice-ml is required.
The easiest way to install recourse-game is using pip:
pip install -U recourse-game
The documentation will include more detailed installation instructions.