Credit card transaction data analysis and performance evaluation of machine learning algorithms for credit card fraud detection

Fraud is defined as any malicious behavior with the intent of causing financial loss to a third party[1]. Due to the unprecedented pandemic scenario prevailing over the world in the past 2 years, there has been rapid participation in online-based transactional activities using credit cards and it has coerced many people into the digital world[2].This increased utilization of advanced digital cash or plastic cash even in new nations has highlighted increased fraud cases related to credit cards[3]. Obviously, this elevated activity has resulted in criminals getting more attracted to it[4]. As part of comprehensive fraud prevention, fraud detection helps to reduce the manual aspects of the screening/checking process and makes the process automatic. This is now one of the most well-known industry/government issues[5]. Detection of potentially fraudulent transactions is the main goal, so as the customers are not charged for the products which were not purchased by them. Additionally, we have to reduce the business costs and make it profitable for credit cards companies. The many strategies used to identify credit card frauds are investigated in this study. This paper examines the algorithms used for detecting fraud related to credit cards, as well as their advantages and disadvantages, along with performance comparisons that will aid in the evaluation of their usability and accuracy.