Publication: Explainable Machine Learning for Real-Time Payment Fraud Detection: Building Trustworthy Models to Protect Financial Transactions
Date
2024
Authors
Al-hchaimi A.A.J.
Alomari M.F.
Muhsen Y.R.
Sulaiman N.B.
Ali S.H.
Journal Title
Journal ISSN
Volume Title
Publisher
Springer Science and Business Media Deutschland GmbH
Abstract
In this study, we introduce an advanced machine learning model integrated with explainable AI techniques to enhance the detection of payment fraud in real-time scenarios within the digital finance sector. As online transactions continue to proliferate, so too do the fraudulent activities associate with them. Our approach effectively differentiates between legitimate and fraudulent transactions by meticulously analyzing key features such as transaction amount, type, and the accounts involved. Through a comprehensive evaluation of various machine learning models, the Decision Tree model emerged as the most effective, achieving an accuracy of 95.4048%, precision of 92.9461%, recall of 98.2456%, and an F1-score of 95.5224%. This study not only proposes a robust and explainable machine learning framework but also significantly enhances the transparency of fraud detection decisions. It equips financial institutions with a potent tool to safeguard their customers? assets against fraud, thereby bolstering the reliability and trustworthiness of digital payment systems. ? The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
Description
Keywords
Crime , Decision trees , E-learning , Machine learning , Network security , AI techniques , Decision-tree model , Digital finance security , Financial transactions , Fraud detection , Machine learning models , Machine-learning , Real- time , Real-time fraud analyse , Transaction analyse , Finance