Reproduce “Fair prediction with disparate impact: A study of bias in recidivism prediction instruments” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Alexandra Chouldechova
The authors analyze COMPAS classifications, fairness measures, and hypothetical sentencing policies. They show that, with different recidivism prevalence across groups, predictive parity generally cannot coexist with equal false-positive and false-negative rates. In Broward County data, high-risk predictive value is similar across Black and White defendants at cutoffs of 4 or higher, while error rates differ; penalty effects depend on policy.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.