Improving E-commerce Fraud Detection via Machine Learning: Comparative Evaluation of Model Effectiveness
T Gangalakshmi, V. Sathya, S. Vinodhini, E. Chandralekha, V. Usha, Siddharth Ravikumar · 2023
In the fast-growing E-commerce world, fraud has become a major concern, causing economic losses and eroding consumer trust. This work aimed to detect E-commerce fraud using machine learning on a customer dataset. The dataset contained transactional and customer data, requiring preprocessing to handle missing values and categorical features. Four models—Logistic Regression, Random Forest, Gradient Boosting, and Support Vector Classifier—were trained and evaluated for fraud detection accuracy using metrics like Accuracy, Precision, Recall, and F1-Score. Results revealed distinct model strengths and weaknesses. Logistic Regression achieved balanced performance, while Random Forest and Gradient Boosting excelled in certain aspects. Support Vector Classifier also demonstrated competitive results. This work highlights the potential of machine learning in E-commerce fraud prevention, enhancing transaction security and trust. Further research could refine and advance fraud detection techniques for online platforms.