A Survival Analysis Methodology for Credit Granting Decision

Student Name: Kamaljeet Bhatia

Instead of classifying credit card customers as defaulter and non-defaulter, the current thinking in credit business is to assess customers as to how profitable they are. The time from a card is issued till the time a customer defaults is an important variable in assessing their profitability. Here an attempt is made to model and predict the survival probability (in terms of time to default) of private label credit card customers for credit granting decisions. The hazard rate of time to default dependent variable is modeled in terms of various demographic and psychographic variables using the Cox proportional hazard model. Significant covariates are identified and survival probabilities are estimated using Breslow's and Kalbfleisch & Prentice's method based on the fitted model, which proved to be extremely useful for the organization where the project was undertaken.