Default of MasterCard Customer Prediction Using Machine Learning Approaches

Abstract

In the era of internet, Use of MasterCard is the best way to pay bills. MasterCard payment is increasing on a large scale day by day. MasterCard allows the convenience of spending on credit to the owner of owner of card. This means that the lack of cash availability at that time is not concern for a MasterCard holder since he/she can spend and purchase on credit and pay conveniently at a later date. Before giving a credit loan to borrowers, bank decides who is bad (defaulter) or good (Non-defaulter) borrower. The prediction of borrower status i.e. in future borrower will be defaulter or non-defaulter is a challenging task for bank. The defaulter prediction is a binary classification problem. There are various existing algorithms for checking the default of MasterCard customer such as Support vector machine, Decision tree, Random forest, linear model etc. As we know defaults of MasterCard customer is increasing day by day so current methods will not be proficient in future, hence there is a need to join two or more methods to improve the performance of present models using ensemble methods. Thus various machine learning models have been explored and analyzed. The models have been evaluated based on various performance metrics. At last models have been compared with existing models to evaluate the accuracy.