A Data-Driven Bilevel Optimization Problem Considering Product Popularity for the E-Commerce Presale Mode

To obtain a competitive advantage in the e-commerce presale mode, an e-tailer needs to design new products favored by consumers and consider the demand matching and supply chain services to achieve better operational results. This paper develops a data-driven optimization model based on consumer analytics and bilevel multiobjective programming to provide practical solutions for the e-commerce presale mode. At the upper level, the e-tailer, as the leader, identifies consumer preferences by quantifying product popularity through consumer analytics. Then, the e-tailer optimizes the product popularity to select the products suitable for presale and formulates the production plan. At the lower level, the logistics enterprise, as the follower, formulates the distribution plan based on the leader's decision. Because consumer analytics are utilized and the model has a bilevel structure, a data-driven optimization method is proposed to conduct simulations for the proposed model. The model uses the multiple objectives binary particle swarm optimization with multiple social structures (MOBGLNPSO) and bilevel multiobjective particle swarm optimization with multiple social structures (Bi-MOGLNPSO). The results analysis and sensitivity analysis verify that the proposed model and method can improve the demand matching and operational efficiency of the e-commerce supply chain and uncover managerial implications for both the e-tailers and logistic enterprises.