A Justice-Based Framework for the Analysis of Algorithmic Fairness-Utility Trade-Offs

This repository is the official implementation of the paper "A Justice-Based Framework for the Analysis of Algorithmic Fairness-Utility Trade-Offs" by Corinna Hertweck,* Joachim Baumann,* Michele Loi and Christoph Heitz (*equal contribution). The paper describes an approach for balancing the utility of the decision maker and the fairness towards the decision subjects for a prediction-based decision-making system. It also includes concepts presented in the paper "Distributive Justice as the Foundational Premise of Fair ML: Unification, Extension, and Interpretation of Group Fairness Metrics" by Joachim Baumann,* Corinna Hertweck,* Michele Loi and Christoph Heitz (*equal contribution), which proposes a general framework for analyzing the fairness of decision systems based on theories of distributive justice and which unifies and extends existing definitions of group fairness criteria.

Requirements

To install requirements, you should have Anaconda installed. Once installed, run:

conda env create -f environment.yml
conda activate utility-based-fairness

Data

The datasets can be found in the folder /data and are taken from Friedler et al. (2019).

Create a different fairness criterion

The Pareto plot shown in Plots.ipynb shows how well a specific fairness criterion is fulfilled. This fairness criterion can be edited by making changes to the config_approach.py. The dataset on which the fairness is measured can be edited in config_data.ipynb. After making such changes, rerun the notebook Plots.ipynb.

Python version used: 3.9.12