Online Payment Fraud Detection Using Machine Learning

Abstract - With the rise of online transactions, payment fraud has become a critical concern. Traditional rule-based systems are no longer effective against advanced fraud tactics. This project uses machine learning algorithms—Logistic Regression, Decision Tree, Random Forest, and Neural Networks—to detect fraud in real time. It addresses the challenge of imbalanced datasets using techniques like SMOTE. Models are evaluated using accuracy, precision, recall, and F1-score, with Random Forest and Neural Networks showing the best results. The system enhances transaction security by accurately identifying suspicious activities, proving the efficiency of machine learning over traditional fraud detection methods. Key Words: Online Payment,Fraud Detection,Machine Learning,Imbalanced Data,SMOTE,Random Forest