Study And Application Of Prediction Model Based On Freight Volume Of Logistics Demands

In recent years, with the development and transformation of the economic situation and structure, the role played by modern logistics industry in the development of national economy and the allocation of resources becomes more and more acute. The logistics industry is regarded as a basic industry in Chinese economy, and its development has always been an indicator which measures the degree of the economic development in a country of region. The national “12th –five-year plan” has push forward accelerating the construction of logistics system and logistics infrastructure construction, reducing logistics costs, improving logistics efficiency, optimizing the layout of the development of logistics industry as well as promoting the logistics intelligence and standardization, and all of these will bring rapid growth of logistics demands. Therefore, it is necessary to establish the reasonable prediction model.This paper will have a deep analysis on the logistics demand prediction model which regards the freight volume as the indicator in the light of development situation of logistics industry. This paper will analyze from the following aspects:1. On the basis of reviewing the basic theory of logistics industry and introducing the research situation both at home and aboard, the paper points out the existing problems in the course of the development of logistics industry and factors which affects the logistics demand. The paper has selected freight volume as the prediction indicator through the method of grey relational analysis and also chose grey prediction model and exponential smoothing model as the basic prediction method to provide the basic information for further improvement of the prediction model.2. According to the existing problems of traditional GM(1,1) model including the difficulties of improvement, limitation of conditions and low prediction accuracy, we reconstruct the background values through quadratic interpolation and the initial values are calculated by moving average method, and the modified GM(1,1) model of logistics demand is established; considering the characteristics of the freight volume, we optimize the smoothing factor and set up the exponential smoothing model based on the minimum of sum of the squared errors; finally, the feasibility and effectiveness of these two improved single models are verified.3. Considering the limitation of these two single models, we have the in-depth study about the combination prediction of the freight volume. First of all, the fixed weight combination prediction model based on the minimum of sum of the squared errors is proposed; furthermore, the improved fixed weight method which combines the vector included angle cosine and ordered weighted geometric average method is established. In order to further improve the accuracy, a variable weight combination prediction model is proposed and applied to solve the weight at any time. At last, the more stable and effective combination prediction models are verified through example simulation.