FRAUD DETECTION IN CARD NOT PRESENT TRANSACTIONS BASED ON BEHAVIORAL PATTERN
Rabiyathul Basariya · 2014
Rapid advancements in technology have imperative effect on consumerism. Technological advancements change the way the world operates. The shifting of customer buying patterns is made viable and sustainable through credit cards. Internet merchants are skeptical on Card Not Present transactions. Every advancement has its own intrinsic worth and frailties. Millions of credit card transactions are processed each day. Increases in online shoppers in turn provide more opportunities for credit card usage, which is directly proportional to commit deception. The dearth of tracking fraudulent credit card transactions is due to the increase in white-collar criminals. The way out to this problem is tracked by the behavioral pattern of the customer by implementing the Classifiers Naive Bayesian and Random Forest to predict the legitimate and fraudulent patterns. The proposed method builds personalize and aggregate model to predict deceitful transactions. Since individual’s transactional behavior varies from one another, there comes the need for personalize and aggregate model. The aggregate model performs better than personalized model. Naive Bayesian approach attains best results for personalized model and Random Forest attains best results for aggregate model.