Research On The Application Of Combined Forecasting Model In Forecasting Of Container Accumulation

Along with the development of forecasting theory, it's popular to use nonlinear forecasting methods. But as a result of the complexity of some forecasting object, limited information resources and sensitivity to hypothesis, traditional individual forecasting models have some shortcomings. In order to spread risk, it's very important role and significance to analyses different combined forecasting models, choose suitable combined forecasting models according to the development of things to improve prediction accuracy. This paper seeks to make a suitable combined forecasting model for the fast growing regional container accumulation, and provide constructive foundation for the development of container transportation.This paper first carries on the summary of the prediction theories and the methods; Then analyzes the influencing factors form macro economy factors such as the conditions of national economy development, the level of international trading, the industrial structure, and the micro influencing factors such as the foreigner trade import and export volume, the rate of suitable box goods of container, and based on the working above ,this paper constructs combined forecasting model of container accumulation; Finally takes the prediction of container accumulation of Hefei as an example, this paper carries on prediction separately by multiple linear regression model, GM(1,1) model, BP neural network model, expansion collection forecasting model four individual models and combined forecasting model, examines and analyzes the accuracy of each forecasting model.