Development of Card Fraud Detection Model with Intelligent Agent Technology and Use of Model Results in Prevention Processes

Changing needs and ease of use are increasing the demand for credit card products in the individual and corporate segments. This tendency towards credit cards brings with it an increase in card fraudster cases. These undesirable situations cause cash and reputational risks, especially in banks. Detection and prevention efforts are being made to prevent fraud cases that are frequently encountered in banks. In the first stage of this paper, in addition to well-known classification (Random Forest and Support Vector Machine) detection models, agent-based models such as Double Deep Q Network have been developed. After the model development process, the outputs of the card fraud detection model were used to create the prevention process. The main purpose of this paper is to use agent technology in card fraud detection and to create a new action flow for prevention efforts. In this way, an agent-supported structure that can detect, evaluate and prevent card fraud behavior has been developed. In addition to the detection model, the study also attempted to support the card fraud management system by developing a prevention process.