Research On Urban Logistics Demand Forecasting Based On Combined Forecasting Method

Logistics demand forecasting as one of the basis of logistics planning, is not only an important premise of designing a variety of logistics development policies and infrastructure construction, but also a key indicator to measure whether the logistics planning is reasonable. In this context, analyzing various socio-economic activity factors'demand satisfaction extent to logistics supply and conducting effective logistics demand forecasting will be propitious for rational planning and construction of logistics infrastructure and to improve the urban logistics supply systems, which will make great significance to enhance the logistics efficiency, reduce logistics costs and realize the sustainable development of China's economy.This thesis will combine related theories and research status to establish an index system of urban logistics demand forecasting, and accordingly propose a logistics demand forecasting method based on combined forecasting models, and finally verify that the combined forecasting model is scientific and effective through an empirical research.Apart from the first chapter of introduction and the last chapter of summary and outlook, the thesis can be mainly divided into four parts. The first part is the review of related theories and forecasting methods of urban logistics demand .The second part will establish an urban logistics forecasting index system based on main influence factors of multiple logistics demand indicators, from the aspect of the characteristics of urban logistics demand, providing momentous premise and basis for the establishment of forecasting models. The third part will propose a combined forecasting model of urban logistics demand based on single-item forecasting models, which needs to select the quadratic programming model targeted at the minimum variance as its objective function from combined forecasting models in the optimal of linearization, then transform it into a constrained least squares model and use MATLAB toolbox to solve its combined weights, which also reduces complex calculation work. Finally, the thesis takes Wuhan city as an example to conduct an empirical study. Firstly use correlation coefficient method and principal components analytic method to conduct pretreatment of indicator data, which greatly simplifies the forecasting index system through reducing the number of input variables and retaining the integrity of data and information. Secondly the establishment of the combined forecasting model and exhaustive contrastive analysis with a variety of single-item forecasting models verify the effectiveness and stability of the combined forecast model, that is, the combined forecasting model can fully integrate the information of each single-item forecasting model and increase the forecasting accuracy, and achieve to reduce the risks of forecasting.