Prediction of churning customers in credit card business

Abstract

The worth of credit card business in US is more than prepaid cards and debit cards, which accounts for 2.5 trillion dollars. Given the heavy competition in this sector, the churn rate has increased in last few year which has put lot of pressures on the managers. So, there is need to predict the churning of customers before they stop using credit card services of that company.one need to understand the various factors which affect this behavior and how these factors are correlated. Based on how likely a customer is going to churn, managers can promote their services and offerings to reduce the churn rate. The project is completed in six steps like a typical marketing research project starting from problem identification to managerial implication. The data was publicly available at the web which has 7000 responses with 24 columns or questionnaire. Data was prepared before performing any analysis. Data analysis was performed to validate the hypothesis developed. Findings:- * Data is skewed towards existing customers as average value of attrition_flag is 0.84. In this dataset, majority of the customers are between 40-50 years of age with almost equal gender split. Credit card holders have 1-3 dependents. * Majority of the credit card holders are either in college or graduates with income in the range of $40K-$80K. Average card utilisation is ~18%. * The credit utilization ratio measures a person's credit card debt compared to their total credit card limits. From the above 2 charts we can clearly see that the card utilization ratio is much more for existing customers than for attired customers. * Majority of the credit card holders are either in college or graduates with income in the range of $40K-$80K. Average card utilisation is ~18%. Managerial Implications:- As a manager of the credit card company, he/she should focus on * Increasing the Utilisation ratio of the card by encouraging people to spend by giving them discounts. * Finding more customers for premium cards as their attrition rate is lower than nonpremium cards. * Cross-sell other products of the company to existing customers to increase the total relationship count. This will lower the attrition rate as evident from the regression equation