This repository contains an analysis of bias mitigation algorithms in machine learning. We apply a range of interventions to two data sets and compare results.
You can view the results of our analysis and an accompanying discussion on our accompanying website.
Notebooks containing all of our analysis can be found in the notebooks/ folder. You can run any one of them in your browser without installing anything courtesy of Binder. Alternatively clone this repository to run the notebooks locally. You can install the required dependencies with
pip install -r ./src/requirements.txtAll source code is contained in src/, including raw versions of the notebooks. See the README in src/ for more details.