Analysis of Techniques for Credit Card Fraud Detection: A Data Mining Perspective

Abstract

The expeditious development of World Wide Web tradition has derived the situation where online shopping and other payment services become most popular among people. They always buy goods and services online or off-line and use their credit or debit card for payment. With the swipe card employment, the fraud rate is also swelling day by day. Hence, the credit card has evolved as the standardized method of payment and the fraud associated with credit card is also expanding exponentially. As we know, many modern data mining techniques have been deployed for the detection of fraud in the domain of credit cards, such as Hidden Markov model, fuzzy logic, K-nearest neighbor, genetic algorithm, Bayesian network, artificial immune system, neural network, decision tree, support vector machine, hybridized method, and ensemble classification. The purpose addressed in this paper is to consolidate various data mining approaches used for finding credit card frauds by researchers to carry out research in the domain and has a state-of-the-art view of the financial domain.