A Method for Add-on Products Recommendation Based on Monetary Preference in RFM Model

This paper draws on the concepts of operations research optimization theory and proposes a diversified add-on product recommendation method to meet the actual needs of consumers who purchase multiple items while adhering to their price preferences. The study employs a multi-stage hybrid approach: first, it predicts product ratings based on consumers' historical purchase data and constructs a candidate product set using association rule algorithms. Second, fuzzy mathematics methods are used to extract consumers' price preference characteristics for different product categories, and the next purchase amount is predicted based on historical purchase behavior. Finally, the recommendation problem is formulated as an optimization model constrained by the predicted purchase amount, and the optimal product combination is obtained using a dynamic programming algorithm. To evaluate the proposed method, experiments are conducted on a data set obtained from Kaggle Big Data Competition. The results show that our method not only keep accuracy and achieves much higher improvement in diversities than the baseline methods, but also could provide add-on product recommendation that meets consumers' single product price preferences and consumption amount preferences. This study innovatively integrates operations research optimization theory with recommendation systems, providing a novel approach to solving add-on product recommendation problems in e-commerce platforms and offering a theoretical foundation for promotion strategy development.