Machine Learning and Credit Counseling

Machine Learning is a technique for analyzing large datasets and applying them to a variety of predictive problems.  Rather than writing explicit models, machine learning systems will progressively “learn” to improve their performance.  Funded by a grant from the Institute of Consumer Money Management, this project aims to use machine learning to analyze the challenges faced by clients in debt management programs and provide insight into how credit counseling agencies can improve their clients’ success rates.  The core model attempts to predict client outcomes in a debt management program based on their financial and demographic characteristics.  Armed with that knowledge, credit counseling agencies can identify clients at risk of dropping out of the program and help them stay on track to achieve financial stability.



  • We have developed the first and only machine learning model for DMP
  • A pilot program is running with participation from six credit counseling agencies
  • The main machine learning model evaluates clients at intake and reports their predicted outcomes and how those outcomes can be improved by interventions
  • An additional model is designed to use missing payment data to predict outcomes for clients who are already in the program