Segmentation of E-Commerce Users on Cart Abandonment and Product Recommendation Using Double Transformer Residual Super-Resolution Networ

A novel approach named “Segmentation of E-commerce users on Cart Abandonment with Product Recommendation using Double Transformer Residual Super-resolution Network (SEC-CAPR-DTRSRN)” is proposed. This study commences with the collection of e-commerce data from a diverse multi-category store. The pre-processing phase uses Fairness-aware Collaborative Filtering (FACF) to suggest personalized items or content to users based on their preferences through behavior. Following pre-processing, the data undergoes segmentation using the Generalized Intuitionistic Fuzzy c-means Clustering (GIFCMC) technique to categorize users to target-based customer groups. To enhance cart transactions, Double Transformer Residual Super-resolution Network (DTRSN) is introduced for product recommendations. The exploration of cart abandonment identifies potential discord stemming from a misalignment between consumer perceptions and digital site experiences, potentially leading to cart abandonment due to user annoyance. While the DTRSN lacks an explicit adoption of optimization systems for calculating optimal parameters, the manuscript proposes the integration of Polar Coordinate Bald Eagle Search Algorithm (PCBESA). PCBESA is introduced to optimize PCBESA DTRSN, ensuring precise product recommendations. The proposed (SEC-CAPR-DTRSRN) method is implemented, and their performances is rigorously evaluated using key metrics, including mean square error, standard deviation, and mean reciprocal rank (MRR). The proposed method gives 12.78%, 29.85%, and 17.45% lower mean square error and 23.67%, 28.86% and 16.45% higher MRR with existing techniques like segmentation of e-commerce users depending on cart abandonment with product recommendation using collaborative filtering, such as moderating the effect of exorbitant pricing (SECU-CAPR-CF), new top-n recommendation technique for multi-criteria collaborative filtering (PR-MCCF-EC) and reinforcement learning e-commerce cart targeting to reduce cart abandonment in e-commerce (RL-EC-RCA) methods, respectively.