Inquiry Response: Improving Fairness and Explainability in Risk Modeling

Inquiry:

We’re concerned about potential bias in the data our risk models use for lending decisions because our data is primarily made up of white males. We’re already working on accuracy with respect to different sub-populations. What are we missing?

Response:

THE BIG IDEAS:

  • Self-fulfilling prophesy
  • Feature impact
  • SHAP and LIME frameworks

SELF-FULFILLING PROPHESY

When we rely exclusively on historical data, our predictions become self-fulfilling prophesies. This means that if you’ve seen many, many examples of success from

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

Log in: