Research on Logistics Demand Forecast based on the Combination of Grey GM (1, 1) and BP Neural Network

Abstract

The steady development of China’s economy has led to the rapid development of the logistics industry. Nowadays, the logistics efficiency in the world has been at a high position, but compared with advanced developed countries, logistics costs are still higher. Establishing an effective logistics demand forecasting model is of great significance to reduce logistics costs and optimize the layout. This paper establishes a combined model based on the research of grey GM (1, 1) and BP neural network considering the scope of application and error. Using the combined model to fit the freight volume data of China in the past 20 years, the results show that the combined model has less error and higher precision in the forecasting of logistics demand.

Regional logistics demand forecasting: a BP neural network approach

“…The BP neural network model has better predictive effect than GM (1,1) model. This finding is consistent with the results of Du and Chen [29] and Wang et al [55]. Scientific and accurate prediction results can provide a reasonable reference for e-commerce platforms, and logistics enterprises to make decisions with the greatest effectiveness.…”

“…The results show that GM (1, 1) model and BP neural network model have a good application prospect in regional logistics demand prediction, and BP neural network model has a relatively small prediction error and a relatively better prediction effect. Meanwhile, the findings of this study are consistent with those of Du and Chen [29], and Wang et al [55], which is helpful to promote the prediction and application of BP neural network model in regional logistics demand, last kilometer logistics demand, crowdsourcing logistics demand and other aspects.…”

“…Before the logistics demand forecast, the establishment of a related indicator system is very important. So far, many researchers (e.g., Nguyen [26]; Fan and Wu [27]; Du and Chen [29]) have paid attention to indicators such as regional GDP, per capita disposable income, total retail sales of social consumer goods, total import and export trade, fixed asset investment in the logistics industry, number of employees in the logistics industry, financial expenditure in logistics and transportation, and mileage of vehicles. However, indicators related to e-commerce are less considered.…”

“…Han et al [28] took GDP, total cost of social logistics, investment in social fixed assets, import, and export volume as indicators of logistics demand calculation and prediction. Du and Chen [29] used GDP, post and telecommunications services, total retail sales of social consumer goods, and residents' consumption level as indicators for logistics demand forecasting. The research of the above scholars has a good reference value, which this paper uses to construct the indicator system of logistics demand prediction for e-commerce development in Guangdong province.…”

“…As the basic, strategic and leading industry of national economic development, logistics industry plays a vital role in the development of regional economy. Based on literature review, this paper follows the following principles when constructing the logistics demand forecasting indicator system: In addition, using the indicators of domestic and foreign scholars such as Ishfaq and Sox [15], Hsiao and Hansen [16], Hsu and Wang [17], Fan and Wu [27], Han et al [28], Du and Chen [29], etc., this study constructed a target layer using the indicators of Guangdong logistics demand prediction. Taking logistics demand environment, commercial trade environment, basic support environment and e-commerce information environment as latent variables, the indicator system takes 13 indicators such as logistics demand scale as observation variables as summarized in Table 2.…”