A Kind Of Economic Time-series Analysis Model And Application

The analysis of time-series is important for economics statistics and forecasting. Up till now, most documents adopt ARIMA model to carry on modeling and predict to time-series analysis extensively. But ARIMA model needs more than 50 historical statistics in model discerning, and it is difficult to collect data by quarter, month or year. If the datum are less than 50, that is to say, datum of necessary modeling are insufficient, model is often relatively poor precision by ARIMA model. And data insufficient accords to grey systematic thought of modeling.But foundation of these grey model requires that the datum are not to be too much, and it is relatively more ideal to about 10. For this reason, using the grey systematic model to datum more in number, This thesis has put forward a kind of new grey systematic modeling method-separated modeling. This method firstly divides the initial data array into two groups properly and carrys on grey modeling to two groups separately; secondly utilizing the moving operator, gets the prediction value of two data arrayswith grey model separately; lastly, adopting proper datum merge way, makes two group prediction data merge together and gets final initial data prediction value.On this foundation, and In the subject - "the macro-economy monitor and early warning system of city of Jinan", the number of data are 24 at most and results received are not too ideal directly by ARIMA model. This text has put forward a kind of new economic time-series model -grey ARIMA model. Let time array has the following formswhere x(t) be determinant trend of w(t) and y(t) be random trend of w(t) . Firstly, utilizing grey-separate model makes time-series take speadily, that is to say, use the grey model to prune the trend one x(t) ; and the array got is a steady time array y(t) ; secondly, using ARIMA models y(t) ; lastly, we get combined predication model of w(t) .This thesis has three parts in all. The first part mainly introduces general modeling course of grey GM(1,1) and ARIMA model. The second is main part, and consists of innovation point of this thesis. It introduces the process of setting up model with grey separated model and ARIMA model. The third part is analysis of real example, and it establishes model and predicts with above-mentioned methods to datum of the subject - "the macro-economy monitor and early warning system of city of Jinan", and can makes the precision of the model raise greatly.