Application Of The Time Series Model To Prediction Of The Consumer Confidence Index

Time series analysis is a method，which utilizes the historical data to discover how thingswith time going. It’s a kind of very effective and useful statistical analysis tool. Supportvector machine regression theory developed by support vector machine classification theorythat based on statistical learning theory. It can effectively solve the problem of machinelearning in the case of small sample situation. Based on the grey theory is a good predictormodel under the condition of incomplete information and less in the statistics.For most of the economic data exist periodic changes and some irregular elements whichbring difficult to research and analysis the future development trend of economic. A huge ofreference shows that the relative error is larger if use one single prediction model. So theforcesting results are not accurate. Thus affecting people’s judgment on current economicsituation. Therefore, searching the reasonable and effective prediction method is necessary.According to this situation, this paper proposes a new hybrid forecasting model. It is basedon the multiplicative seasonal ARIMA model, support vector regression model (SVR) andGM(1,1) model. We use the consumer confidence index of China to exanine the predictionaccuracy of the hybrid model. The prediction performance was compared among thesemethods, i.e., the multiplicative seasonal ARIMA model,Support vector regression model,GM(1,1) model and the hybrid model.(All models use one step prediction in this article). Theresults shown that,the MSE and MAPE of the hybrid forecasting model were the lowest.