Comparative Evaluation of Machine Learning Algorithms for Credit Card Fraud Detection

Banks are the backbone of the economy of any country, and hence, it is essential that they function properly so as to achieve a booming and sustainable economy. The major barrier for banks is the defaults on the credits they give to the customer. As banks profit mostly from the loans they give to customer, hence, it is required that they possess a solid and efficient model to minimize the losses they incur via the credit defaults. This article assesses the credit card default prediction's performance. Earlier the credit card defaults were tallied using standard tools such as FICO scores, but with the development of machine learning, it became much easier to build highly effective risk prediction model. Anyone with a bit of machine learning knowledge would know that credit risk prediction is nothing but a mere binary classification problem. Thus, various machine learning models such as KNN, decision tree, random forest, and logistics regression have been applied. When we look into the credit risk of credit card clients, it indicates that random forest best specifies the aspects to examine with an accuracy of 82% and an area under curve of roughly 77%.