Personalized Fraud Detection in Online Banking Using Sequential Pattern Mining and Machine Learning

Abstract

With the rapid expansion of online banking and the increase in financial transactions on digital platforms, fraud detection in these systems has become a major challenge for financial institutions. Traditional fraud detection methods, which are typically based on predefined rules, are unable to cope with new threats due to the complexity and diversity of frauds and the rapid changes in user behavior. In this paper, a personalized approach for fraud detection in online banking using sequential pattern mining and machine learning is proposed. In this approach, each user's financial transactions are analyzed as time sequences to identify the normal behavioral patterns of each user and extract suspicious deviations. Advanced algorithms like PrefixSpan are used for extracting sequential patterns, and machine learning algorithms such as Random Forest and SVM are employed to classify transactions as fraudulent or non-fraudulent. Experimental results show that the proposed model outperforms traditional methods, with higher accuracy and better ability to detect complex frauds. This research can contribute to improving fraud detection systems in online banking and enhancing the security of financial transactions.