Predicting Customers Intending To Cancel Credit Card Subscriptions Using Machine Learning Algorithms: A Case Study

Abstract

Following paper introduces analysis of machine learning algorithms implemented in order to predict customers of commercial bank who may be in risk of cancelling credit card subscriptions by following three months after a year or less activity. An analysis of various data preprocessing, sampling and structuring procedures using a feature set made up of 106 variables -describing customers’ transaction activity, demographics, overall contentment and relative information to consumer experience-also shared. Study also includes performance comparison of Deep Neural Networks against other generic machine learning algorithms on two different cases. Deep Neural Networks were the point of interest of this study and it turns out, them to perform better than generic machine learning algorithms.