Framework of Fraudulent Credit Card Preventing System Using Machine Learning Techniques in Vietnam

Online payments have been popular for consumers, particularly in coronavirus disease 2019 (COVID-19) pandemic time in Vietnam. However, there exist some certain limitations of using online payment in Vietnam such as high card issued fees, online transactions but paying with cash, complicated interbank payment services, or lack of card payment of points. At the moment, not many financial and credit institutions in Vietnam have installed automatically fraudulent preventing system to protect their customers. Therefore, building a framework of transaction warning systems using machine learning techniques is challenged for many researchers. The survey on Vietnamese banks’ customers has shown the issues of making online payments. The most of their concerns was unsafe processing problem. Therefore, having an automatically warning system will be a great supporting system for the online customers. In online payments areas, machine learning can help bank staffs to prevent unusual transactions such as transaction of buying/selling from the same account with alternative unfamiliar geographical location, trade with large amounts in many times or with the same newly opened account, etc. The machine learning has been applied through filtering and analyzing the black list accounts’ transactions which had the same characteristics. In this research, we propose a framework of fraudulent credit card preventing system in where it uses machine learning techniques to define unusual transactions. A focusing on algorithm of balancing process is proposed as the most challenged issue happening in machine learning process. The balanced dataset experiments show the higher accuracy than the one without using balancing algorithms.