Credit Card Transaction Monitoring using Machine Learning Techniques

Abstract—Credit card fraud is a significant worry for financial institutions and cardholders globally. As digital transactions have become more prevalent, the risk and complexity of fraud have also grown. This paper compares different ML and DL algorithms for detecting Online fraudulent activities. The model leverages real-time transaction data and behavioral patterns such as location, amount, and time to detect anomalies. We created a synthetic dataset replicating real-world scenarios and evaluated models like Logistic Regression, Random Forest, XGBoost, and LSTM. Our results show that deep learning techniques, especially LSTM, outperform traditional methods in detecting fraud with high accuracy and low false positive rates. Index Terms— Online Fraud, Machine Learning, Deep Learning, Anomaly Detection, LSTM, Random Forest, Predictive Modeling.