The implementation and effectiveness of fairness toolkits in machine learning, specifically AIF360 and Fairlearn, significantly differ in terms of ease of integration, data handling capabilities, metric variety, and algorithmic performance. These differences can substantially impact their adoption and effectiveness in various socio- economic and organizational contexts, influencing the overall fairness of machine learning systems in real-world applications. This work aims to evaluate the differences between those most popular fairness toolkits.

This project requires Python 3.10. If you are using pyenv, the Python version will be set automatically by the .python-version file.