Fuzzy-Based Fraud Prediction System in Digital Transaction

Growth in digital transactions is observed over a long period of time due to convenience and time savvy services provided by various emerging technologies and digital platforms. Numerous fraudulent complaints in digital transactions are observed despite of existence of various security mechanisms and tools. Fraudsters are constantly adapting innovative techniques and measures to scale up their fraudulent activities over the Internet. Therefore, it is essential to evolve novel techniques that can help in fraud detection and which motivates our research. This research work aims to develop a fraud detection system that predicts occurrence of fraud specifically in digital transactions. This research paper proposes two algorithms CPFEA and FPA where the former captures the characteristic feature from the customer profile and the latter identifies the fraud score of the transaction. Fuzzy logic plays an important role to determine the probability of a transaction to be classified as a fraud in three levels namely low, medium, and high based on the fraud score for each requested transaction. The performance of the proposed work is assessed using standard measures and the proposed approach has shown satisfactory performance in comparison to some existing algorithms.