Portfolio / Credit underwriting model
Credit underwriting model
An ML ensemble that replaced a rule-based loan approval system and improved accuracy by 18%.
+18%accuracy over the rule-based system
Ensembleof ML models
Power BIdashboards for the business
The problem
Fixed approval rules were rejecting good applicants and approving risky ones, and the business team had no clear view of performance.
What I built
- An ML ensemble model to replace the rule-based system
- Evaluation against the existing rules on historical decisions
- Power BI dashboards for the business team
The result
Better approval decisions, and a business team that can see exactly how the model performs.