Multi-Perspective Fraud Detection Method For Multi Participant E-Commerce Transactions Using Machine Learning

Abstract

The growth of online shopping platforms has significantly increased the number of digital transactions, which has also led to a rise in fraudulent activities. Detecting such fraud efficiently has become an essential requirement for maintaining secure and trustworthy e-commerce systems. This project presents a machine learning-based approach for identifying fraudulent transactions by analyzing historical transaction data and user behavior patterns. The proposed system follows a structured workflow that includes data cleaning, feature selection, handling of imbalanced data, and model training. Supervised learning algorithms such as Logistic Regression, Random Forest, and XGBoost are applied to classify transactions as genuine or fraudulent. The performance of these models is evaluated using multiple metrics to ensure accurate and reliable detection. The results demonstrate that machine learning techniques can effectively capture complex fraud patterns and outperform traditional rule-based systems. This framework can be integrated into real-time e-commerce platforms to enhance transaction security and reduce financial risk.