Fraud Detection and Personalized Recommendations on Synthetic E-Commerce Data with ML

Abstract

The phenomenal growth in the e comm ecosystem has further increased the opportunity for digital trade and also the risk of fake transactions. The integration of artificial intelligence (AI) and machine learning (ML) for fraud detection and personalized recommendations is one data-driven approach for enhancing customer trust and security of operations. This paper investigates the development and evaluation of Model Learning (ML) - based models using synthetic e-commerce data sets to identify fraudulent behaviors in e commerce in conjunction with increasing the accuracy of recommendations. Drawing upon research findings in the literature on the use of AI in identity verification (Alim et al., 2020) and behavioral analytics in trading systems (Zhao et al., 2019), the proposed framework uses big data analytics (Nwaimo et al., 2019; Hwang & Chen, 2017) and predictive modelling (Dugbartey, 2019) to identify anomalous patterns in online transactions. The synthetic dataset reproduces consumer activities in the context of the e-commerce sector, thereby creating a controlled setting for experimentation for algorithms that bypasses privacy violations. Furthermore, personalization methods are formulated under reinforcement learning and sentiment analysis (Rahat et al., 2020; Zhao et al., 2019) to match product recommendations with user preferences. Empirical evidence suggests that hybrid learning models outperform traditional classification methods on the precision/recall metrics for fraud detection, while also improving customer engagement and uplift by adapting to the customers' behavior with the help of adaptive recommendations. This research adds value to the rising research on the development of intelligent e-commerce securities and decision support systems (Olsak & Zurada, 2020; Ali et al., 2019) for informing the development of innovative digital businesses in emerging digital markets.