Credit Card Fraud Detection System Using Machine Learning

Abstract

The rapid growth of online payment systems has made credit cards a widely used more of digital transactions. Along with this growth, incidents of credit card fraud have increased, leading to financial losses for banks and reduced confidence among customers. Conventional fraud detection approaches are mostly based on predefined rules, which are often ineffective in identifying newly emerging fraud patterns, particularly when fraudulent transactions form only a small portion of the overall data. To address this challenge, this study applies machine learning techniques for identifying fraudulent credit card transactions. A major issue in this domain is the highly imbalanced nature of transaction data, where legitimate transaction significantly outnumber fraudulent ones. To reduce the impact of data imbalance, the Synthetic Minority Over-Sampling Technique (SMOTE) is used to generate additional fraud samples and enhance model training. Multiple machine learning classifiers are developed to learn transaction patterns and distinguish between genuine and fraudulent activities. Model performance is assessed using recall, F1-Score, and the Area under the Precision-Recall Curve (AUPRC), as these metrics provide a more reliable evaluation for imbalanced datasets. Experimental analysis indicates that the Random Forest classifier outperforms other models, achieving an accuracy of 99.95%. The findings demonstrate that machine learning-based approaches can significantly enhance fraud detection systems and support effective real-time monitoring of credit card transactions.