Aequitas is an open-source Python library designed for detecting and mitigating bias in data, algorithms, and social contexts. It provides a comprehensive set of tools for developers and stakeholders to assess, repair, and design AI systems with fairness in mind. The library supports state-of-the-art bias mitigation techniques and is fully compatible with the existing Python data science ecosystem.
Overview:
.
├── aequitas # main package
│ ├── detection/ # bias detection tools
│ ├── gateway # remote interaction with other services
│ │ ├── aequitas/ # Aequitas API
│ │ └── AIoD/ # AIoD API
│ ├── mitigation # bias mitigation tools
│ │ ├── data.py # pre-processing tools
│ │ └── models.py # in-processing tools
│ └── tools
│ └── data_manip.py # utilities for data manipulation
├── datasets/ # datasets used for examples
├── examples/ # examples of usage of the library (JuPyter notebooks)
├── LICENSE # Apache 2.0 License file
├── package.json
├── package-lock.json
├── pyproject.toml
├── README.md
├── release.config.js
├── renovate.json
├── requirements-dev.txt
├── requirements.txt
└── setup.pyTo install the library, you can use pip:
git clone https://github.com/aequitas-aod/aequitas-lib
cd aequitas-lib
python -m venv .venv
source .venv/bin/activate
pip install -r requirements-dev.txt
# play with examples in the examples folder