Enforcing fairness in binary and multiclass classification

implementation of fair dummies

Supervised link prediction in a social network using graph-based features, Gradient Boosting, and fairness evaluation with Fairlearn.

This repository was used for my thesis. The goal was to find a biased dataset, and mitigate its bias. That is done under the patients directory. Check the README file for more.

Static fairness audit bench for a real UCI Adult classifier: 6 group-fairness criteria, an EEOC four-fifths check, and a live accuracy-versus-fairness Pareto frontier, no LLM, no API key