Journal of Engineering Research and Reports

This study investigates the integration of cloud-based Artificial Intelligence (AI) solutions for real-time compliance and risk monitoring in e-commerce platforms, addressing the pressing need for scalable oversight as transaction volumes and complexities surge. Utilizing the Olist Brazilian E-commerce dataset, which captures diverse vendor behaviors representative of both large and small enterprises, the research adopts a multi-phase quantitative approach. Descriptive statistics were used to establish baseline transactional patterns, while anomaly detection via Isolation Forest identified outliers indicative of potential compliance breaches. Logistic regression was employed to evaluate the relationship between vendor behaviors such as delivery delays, order cancellations, and product quantities and customer dissatisfaction, illuminating the multifactorial nature of compliance risks that cannot be attributed to single variables alone. Principal Component Analysis (PCA) distilled critical factors essential for effective monitoring. To ensure the robustness and generalizability of the proposed AI framework, 10-fold cross-validation was conducted, achieving an average predictive accuracy of 95.3%. Expert validation through Likert-scale surveys further confirmed the practical feasibility of the framework. The study demonstrates that combining anomaly detection, behavioral modeling, and dimensionality reduction within a cloud-based infrastructure enables dynamic and reliable compliance oversight. Recommendations include standardizing AI integration guidelines, strengthening workforce skills in AI deployment, and reinforcing data governance frameworks to promote scalable, secure, and effective compliance monitoring in rapidly evolving e-commerce environments.