Application Of The Combination Model In Analyzing And Forecasting The Economic Time Series

Time series analysis and forecasting are based on the assumption that the things of thepast will continue into the future and the real thing is the result of the historical development,and things in the future is also the realistic extension under the premise that trends willinfluent the future changes. How to make full use of the past time series to find out the way inwhich past dates influent the present and future date is the important task of the establishmentof the time series model.Now, economic time series analysis and forecasting has been usedwidely. How to make better use of the effective models to find out the operation mechanismof the history data has always been the goal of the research.In forecasting practice, using only one prediction method to get accurate and reliablepredictions is difficult. Different forecasting methods provide different information, ifabandon simply some forecasting methods of huge error will lead to the waste of some usefulinformation. If we get the appropriate combination of the individual model, we can effectivelyuse the single model, making up for the individual model, So as to make full use of theexisting information resources, we attempt to get the proper combination models.Autoregressive integrated moving average (ARIMA) and Support Vector Machine(SVM) are two important and effective analysis and prediction methods of time series. Theycan to some extent reflect the information that the data contains, and the information are notcompletely overlapped. To take advantage of the unique strength of ARIMA and SVR modelsin linear and nonlinear modeling,this paper use the combination mode of this two methods toanalyze and forecast the time series of consumer price index (CPI) and Shanghai compositeindex. Comparing with the result of the single method, the combined mode indicates that itimprove forecasting accuracy achieved by either of the models used separately.