Article Preview
TopIntroduction
The development of e-commerce is based on the popularization of the Internet. By fully using the fast Internet and the security and reliability of network support, it realizes the rapid interaction of information flow and capital flow with consumers and reaches the transaction intention. Finally, the physical goods are delivered to consumers by express and other logistics methods, and the consumers confirm the receipt and complete the final transaction (Wu et al., 2017). Among the three flows of capital flow, information flow and logistics, logistics has the most significant time delay. Under the Business to Consumer (B2C) e-commerce mode, the achievement rate of e-commerce to consumer demand directly determines whether it can gain an advantage in the competition. The interaction and transfer of information flow and capital flow can be completed instantly, but the delivery of physical goods must beachieved through logistics. Therefore, under the B2C e-commerce mode, logistics has become a vital link affecting the competitiveness of e-commerce enterprises. B2C e-commerce platform needs to purchase, store, conduct warehouse management, packaging, circulation, processing and distribution of goods, and finally deliver the goods to consumers to complete the completion of a transaction (Zhang, 2020; Abdulkareem et al., 2021). While the business scale of B2C e-commerce is expanding rapidly, it is not impossible to set up a general warehouse to ensure high logistics service quality, and all requirements are delivered from the general warehouse. It is necessary to set up sub-center warehouses in the areas involved in the business to ensure high logistics efficiency and service quality (Biagi and Falk, 2017; Jannach et al., 2017; Vinodhini and Chandrasekaran, 2017).
The e-commerce platform based on B2C mode can conduct direct transactions with consumers, greatly reducing the level of intermediate retailers and effectively reducing the “bullwhip effect”, but this does not mean that the change information of consumer demand can be easily obtained. In particular, in the modern era of the short life cycle of consumer goods and rapid market change, consumers' demand is affected by multiple factors. It fluctuates wildly, bringing certain pressure and challenges to the inventory management of e-commerce (Xiong et al., 2020). Li and Huang (2019) pointed out that data mining technology refers to using computers as tools. The statistical learning model is constructed and applied to “new” data for prediction and analysis based on data. Its essence is consistent with statistical and machine learning (Li and Huang, 2019). However, e-commerce has the advantage of big data, which can mine consumers' consumption records, transaction browsing behavior and other data. Shahrel et al. (2021) explored the application of the time series model in e-commerce platforms, and the final error is less than 20%. However, there is an apparent deficiency in its research, that is, the time series model only analyzes the historical demand data of consumers, while the multi-dimensional data characteristics of consumers have not been fully utilized. Besides, the predicted value of this research method also lags in time (Shahrel et al., 2021; Tabassum et al., 2022; Savoli & Bhatt, 2022; Ngassam et al., 2022).