An Analytical Approach to Fraudulent Credit Card Transaction Detection using Various Machine Learning Algorithms

Abstract

Technology and the revolution in communication have increased the popularity of digital money usage. Most of the monetary transactions currently take place digitally. It is more convenient and increases the ease for the user. But one major problem in digital money and credit card usage is security. With the increase in credit card usage, security issues increase correspondingly. Many studies and research work are going on to avoid and prevent such practices from taking place. Moreover, various studies on real-international credit scorecard statistics are attributable to confidentiality issues. This paper focuses on current credit card fraud practices and fraud detection methods implemented in real time. Different ML algorithms like fuzzy-based SVM (FSVM), random forest (RF), logistic regression (LR), and support vector machine (SVM) for fraudulent transaction detection on the dataset collected from credit card users have been used to classify legitimate and fraudulent transactions. The comparative analysis of the credit card fraud detection scheme using these classification models was performed with precision, accuracy, sensitivity, and specificity. The comparative analysis outcomes showed that the highest performance was given by the FS VM over other algorithms with an accuracy of 98.61%.