CardSim: A Bayesian Simulator for Payment Card Fraud Detection Research

Payment fraud has been high in recent years, and as criminals gain access to capability-enhancing generative AI tools, there is a growing need for innovative fraud detection research. However, the pace, diversity, and reproducibility of such research are inhibited by the dearth of publicly available payment transaction data. A few payment simulation methodologies have been developed to help narrow the payment transaction data gap without compromising important data privacy and security expectations. While these simulation approaches have enabled research advancements, more work is needed to generate datasets that reflect diverse and evolving fraud tactics. This paper introduces CardSim, a flexible, scalable payment card transaction simulation methodology that extends the small but emerging body of simulators available for payment fraud modeling research. CardSim is novel in the extent to which it is calibrated to publicly available data and in its Bayesian approach to associating payment transaction features with fraud. The simulator’s modular structure, which is operationalized in a corresponding software package, makes it easy to update based on new evidence about payment trends or fraud patterns. After laying out the simulation methodology, I show how outputs can be used to test and evaluate machine learning workflows, modeling approaches, and interpretability frameworks that are relevant for payment card fraud detection.