Statistical and Spending Behavior based Fraud Detection of Card-based Payment System

Abstract

The occurrence of credit and debit card fraud has been a growing issue in the past years, not least due to the increased prevalence of card payments. This results in billions of dollars of lost revenue every year. The fraud preventative measures still show room for improvement. In this paper, we present an alternative fraud detection technique in the form of a rudimentary fraud detection system that utilizes consumer spending behavior. Three attributes of a transaction, namely time, amount, and geographical location, were used as a basis to build a consumer profile. Data for these attributes would be collected from each transaction made by the cardholder and would be used to calculate various statistical values pertaining to their spending patterns, which is used to calculate the probability of fraud. Experimental results show that the aspect of consumer spending behavior can be quantified and used to accurately calculate various probability values related to transactions and possibility of fraud.