The Decision of Product Market Forecasting and Delisting Based on PSO-BP

Abstract

Objective: Aiming to construct a product sales forecasting model with higher forecasting accuracy to make guidance on product delisting and maximize product sales revenue. Methods: This paper uses the particle swarm optimization (PSO) algorithm to optimize the back propagation (BP) neural network of the sales model, and selects the iPhone 11 cell phone sale record as the data set for experimental simulation of the sales forecasting model. The delisting model makes forecasting on the sales revenue of the same cell phone with different memory in a future period. A reasonable product delisting time scale is set, and the delisting time point is determined in conjunction with the revenue situation. Results: The empirical results show that all evaluation indexes of the forecasting accuracy of the PSO-BP neural network forecasting method used in this paper can be improved by more than 20%. Meanwhile, when cell phones with a different memory of the same model are sold in the same market, the cell phones with the medium level of memory have a longer life cycle than those with large and small memory, and the total revenue obtained from selling the cell phones with the middle level of memory is higher. Conclusion: Some insights can be drawn from the study. The PSO-BP neural network forecasting method not only has higher forecasting accuracy than the BP network, but also has better generalization performance and robustness. Since the lifecycle of cell phones with medium memory and the revenue obtained from sales are higher, companies should produce more cell phones with that memory size to maximize corporate revenue. The forecasting results provide a powerful guide for enterprises to make a reasonable dynamic adjustment. These insights of this paper provide theoretical guidance for enterprises to make product retirement decisions.