The Combination Forecast Model Based On Times Series Moedl

Time series analysis provides a dynamic data processing method with scientific basis,following the basic principles as a professional branch in applied fields of probability andmathematical statistics. The method adopts appropriate mathematical model toapproximate describe the various types of data. So as to forecast the trend of the datadevelopment, it needs to analysis the model to essentially understand the inner structureand complex characteristics of data. The prediction is to infer and measure the futuredevelopment and change of things, and economic prediction utilizes the scientific methodto conjecture the variation tendency of objective economic process based on certain timeseries and according to the objective laws of economic. In order to accurately predict, itwould hope fitting accuracy and forecast accuracy of model to be best as far as possible.Because any single forecasting method is difficult to obtain the very satisfactoryprediction results, so people combine all kinds of single forecasting methods to make fulluse of their advantages and to achieve the optimal forecasting results. The paper studiesthe combination forecast model and its applications based on time series model.First of all, this paper introduces the application prospect and the present researchstatus of combination forecast model, the source of the time series analysis and researchand development at home and abroad, the grief history of artificial neural network, thedynamic development of the grey theory and summarizes the research purpose,significance and content of this topic. Then, the paper introduces related conclusions ofARIMA model and the basic principle of model identification, and gives the way ofdealing how to be steady of the time series.Secondly, the paper comes up with a new combination forecast model which isconsisted of ARIMA model, auto-regression model and the BP neural network model withthe suitable method, and makes use of the per capita GDP data of Shanxi province topredict.At last, this paper uses of ARIMA model to identify and fit time series data, then withthe gray model and optimized GM(1,1) model to fitting and prediction, and finally it obtains combination forecast model with ARIMA model and optimized GM(1,1) model bygiven reasonable weight through these methods. And an example is given to proving.