Our new open-source Python library makes fairness evaluation accessible to every ML team.
Today we are releasing FairScore — a zero-dependency Python library that adds five fairness metrics to any scikit-learn-compatible model in three lines of code.
from fairscore import audit\nreport = audit(model, X_test, y_test, sensitive_col="gender")\nreport.summary()FairScore computes demographic parity, equalised odds, individual fairness, calibration by group, and a composite "FairScore" (0–100). It outputs a human-readable report and a machine-readable JSON artefact ready to commit to your model card.
Star us on GitHub and help us reach 1,000 stars this month!