Machine learning models for predictive underwriting
Case Studies / FinTech / Customer Acquisition
Large West Coast based Fintech was looking to expand lending to High Risk Prospects.
Our goal was to leverage ML based modeling techniques to demonstrate increased Risk Differentiation. Also there needed to be an easy way to describe decline reasons at the Prospect level.
Unique features (e.g. with credit tradeline data) and appropriate Reject Inference methods were incorporated in model build. Proprietary Reason Explainer to identify adverse action reason codes in real-time.
Delivered a 5%+ increase in AUC – leading to 70% increase in approved population at similar level of portfolio risk.