The client, RedRock Financial, LLC, underwrites loans for used cars. They contracted us to computerize their underwriting process and build machine learning models to score their loan applicants.
We developed a web-based system for the client, including the database, servers and user interface. We built logistic regression and XGBoost machine learning models to predict the default risk using client’s proprietary datasets. We also created visualizations to show important features and their effects on default risk.
Using Python and R, we created a machine learning powered web 2.0 application that allows the client to input their applicants’ data and get back model predicted default risk instantaneously. Our models have stood the test of time, and the client uses them to this day.