Data-Driven Visual Analytics for Fraud Pattern Discovery in E-Commerce Systems

E-commerce platforms are increasingly vulnerable to fraudulent transactions, necessitating robust and interpretable analytical approaches for timely detection. This study investigates patterns of fraudulent behavior using a publicly available ecommerce fraud dataset by integrating Python-based data preprocessing with Tableau-driven visual analytics. Transactional, behavioral, verification, and geographical features are systematically analyzed through interactive dashboards and statistical visualizations. The findings reveal strong associations between fraudulent activity and key indicators, including elevated transaction amounts, short account lifespans, verification failures, extended shipping distances, and inconsistencies between billing and shipping countries. The results demonstrate that visual analytics not only enhances the interpretability of complex fraud related patterns but also supports hypothesis-driven exploration, offering practical insights for the design of real-time and decision support-oriented fraud detection systems.