Launches
Fairlearn 0.15 fixes a bug that could hide model fairness gaps
Source-backed 2 sources
F
Fairlearn
Tool Oct 12, 2026
A Python library for assessing and improving the fairness of machine-learning models; version 0.15.0 adds group-level plots and fixes weighted fairness calculations.
Visit primary link
2 sources attached
What happened lately
Fairlearn timeline
Fairlearn provides Python tools for measuring and mitigating disparities in machine-learning systems. Version 0.15.0 adds ROC curves split by sensitive group, automatically plots available bootstrap confidence intervals, and corrects several calculation and preprocessing errors. One fix addresses weighted groups with a single sample that could cause demographic-parity metrics to report no disparity despite different selection rates. The release requires Python 3.11 or newer and scikit-learn 1.6 or newer.