With the rapid growth of the e-commerce business scale, to meet customers' demand for efficient order processing, it is of great significance to establish an order management mechanism capable of responding quickly by accurately predicting product demand. This study used real e-commerce order demand data and established a nonlinear autoregressive neural network (NAR) model after pre-processing methods including down-sampling and data set partition to effectively forecast the demand of products in the next 13 weeks. Compared with the Prophet time series prediction framework, NAR had better generalization ability, and the prediction time was reduced by 18.54%. Finally, we summarized two methods' characteristics and gave instructions on applying our model in the real scene. After being deployed in the actual demand management, the trained artificial neural network provides a scientific reference for the data-driven e-commerce decision-making process and brings new advantages over other companies, achieving the rational allocation of resources.

Establishing a rapid-response mechanism to manage customer orders is very important in managing demand surges. In this study, combined with predicting order requests, we established a multiobjective optimization model to solve the warehouse space allocation problem. First, we developed a model based on the NAR neural network to predict order requests. Subsequently, we used the improved NSGA-III based on good point set theory to construct a multiobjective optimization model to minimize resource loss, maximize efficiency in goods selection, and maximize goods accumulation. The following three modes were tested to allocate warehouse storage space: random, ABC, and prediction-oriented. Finally, using actual order data, we conducted a comparative analysis of the three modes regarding their efficiency in goods selection. The method proposed by this study improved goods selection efficiency by a sizable margin (23.8%).