Bibliometric Analysis and Visualization of Machine Learning-Based Credit Card Fraud Detection

Machine learning is often used in credit card fraud detection. Its ability to analyze large amounts of transaction data and identify patterns of fraudulent activity. The aim of this study is to provide insights into the research status, mapping process and annual topics of machine learning research on credit card fraud detection, and to provide relevant references for future research. Research phases with three approaches. The first approach is descriptive statistical analysis for data collection using Scopus web and Herzing’s Publish or Perish. The second approach uses quantitative methods for citation analysis with a bibliometric approach using Vos Viewer application. The research findings are related to the analysis of credit card fraud research divided into three clusters, the first cluster is the research stream using machine learning in data mining, the second cluster is the research stream using machine learning to detect credit card fraud, the third cluster is the research stream using machine learning to classify credit card fraud. The second cluster can resolve the complexity of credit card transaction data and increase the accuracy of the fraud detection system, for the third cluster, we can build a more effective classification in resolving imbalanced data and limited transaction records. A growing research trend is in the third cluster, research related to the performance of credit card fraud classification based on unbalanced data and limited fraud data. Further research is recommended to examine the evolution of the literature on the use of machine learning to detect credit card fraud and to conduct comparative studies between the Scopus, Web of Science and ScienceDirect databases to expand the literature.