Inquiry Response: Notes on Assessing Decision-Making Models for Bias

Inquiry:

We build models in-house models to determine lending for people who buy our products. We rely on these algorithms for auto-approvals, and we’re concerned about model bias. How can we address this issue?

Response:

THE BIG IDEAS:

  • Sub-populations
  • Disparate impact
  • Demographic data

SUB-POPULATIONS

Regulations for bias in modeling are coming. Be prepared by assessing how well you measure model accuracy. For lending, in particular, it’s important to understand whether accuracy differs for particular sub-populations.

  • Your

Become a Research & Advisory Network Client to see the rest. Contact us to get started

Log in: