Risk Scoring for Low Income Applicants

Risk Scoring for Applicants with Low Income and Incomplete Files
Average: 5 (2 votes)
Number of Collaborators
Project Status
Project Type
Project File
Project Description

Used latent measurement models and logistic regression in combination with decision trees to develop predictive models of loan default, and then risk scores. Developed computing algorithms in R, then worked together with engineers in incorporating R codes into Ruby based web decision making systems.

The system got deployed into a company's loan decision making that dramatically reduced its default rate.


RM4E data flow

Jidong Li
Upload Workflow
Graph09483.93 KB
Workflow Description
General RM4E workflow for data analysis which includes the Equation (model architecture), Estimation, Evaluation and Execution parts.