Performance Analytics of Online Payment Fraud Detection Framework

Cybercriminals, fraudsters, and hackers are using the online payment transaction system as a deadly weapon posing a significant threat to customers in the form of online payment transaction fraud. Cybercriminals exploit user vulnerabilities using sophisticated techniques, including advanced digital technology, machine learning, and artificial intelligence to perform fraudulent transactions. The aim of this study is to upgrade the security of the Existing Online Payment Transaction System (EOTS) with a focus on protecting end users and significantly mitigating fraudulent transactions to a great extent. To achieve this aim, a Three-Defense Wall Authentication Framework (TDWAF) that works at the authentication level, a Fraud Detection Decision Support System Framework (FDDSSF) that works at the fraud detection level, and a Deep Authentication Fraud Detection Model (DAFDM) that works at the fraud detection level have presented. In this context, the verification of the authenticity of the two frameworks and one model has been shown in this chapter. Standard performance metrics have been used to test the accuracy, precision, specificity, F1-score, G-mean, and AUC-ROC, and two different datasets have been used. The accuracy results of the proposed frameworks and model have shown remarkable achievements in mitigating fraudulent transactions.