Modelling Credit Card Fraud Data using Machine Learning Algorithms

Abstract

Credit card fraud refers to the unauthorized use of a credit card, often for illegitimate or illegal transactions. In recent years, it has emerged as a major concern, causing billions of dollars in losses annually, according to statistics. Moreover, the problem is becoming increasingly complex with the development of new fraud techniques. This alarming statistic underscores the urgent need for robust statistical analysis to understand, prevent, and combat fraudulent activities using credit card fraud data generated by European credit cardholders. Therefore, employing machine learning models with high accuracy ratings and optimal performance is essential for detecting credit card fraud. This study uses supervised machine learning techniques; decision trees (DT), Random Forests (RF), Artificial Neural Networks (ANN), Naïve Bayes (NB), and Logistic Regression (LR) to detect credit card fraud. The findings reveal that while identity theft, skimming, counterfeit cards, mail intercept fraud, and lost or stolen cards remain prevalent, there is a notable increase in other forms of fraud due to evolving techniques. Among the machine learning models evaluated, the Decision Tree method demonstrated the highest accuracy, outperforming the others.