Audit: Tenant Screening / Rental Application Bias

[yakew7]

Summary

New audit in an untouched domain: tenant-screening models that score rental applicants. Follow the standard pipeline — train biased model → measure fairness gap → identify proxies → remove → retrain → measure.

Why

Automated tenant screening (eviction history, credit, criminal records) is a documented source of housing discrimination and a fresh domain alongside the existing six audits.

Tasks

• Public dataset (open an issue comment with the source first, per CONTRIBUTING §1)
• Tenant Screening/unfair.py + fair.py matching the folder layout in CONTRIBUTING §2
• Identify proxies (eviction history, criminal record as proxies for race); before/after PNGs
• Notebook + README entry