top of page


Predicting Missed Payments from Credit Card Clients

Song Gao, Juergen Kritschgau


Credit card clients miss payments on their credit card debt for a variety of reasons. Being able to predict missed payments would allow banks, credit raters, and debt collectors to forecast their own operations, target interventions or financial products, and accurately appraise the value of credit card debt. In this project, we attempt to use a client’s payment history over a 6 month and demographic information to predict whether the client will miss a credit card payment next month. We used data obtained from a Kaggle competition page to train different classification models, including logistic regressions, Bayesian models, Support Vector Classifiers, K-nearest neighbor models, and decision trees. We used cross-validation and classification accuracy to compare different classification models. Our primary finding is that no model is able to accurately predict whether or not a client will miss a payment.

Screen Shot 2022-06-03 at 11.31.35 AM.png
github URL
bottom of page