Applying Study On Time Series Modeling And Model Selection

Time series analysis is the popularity of economics and statistics of research. People establish maths model by means of analysze actual time data so as to forecast the future of the development. Box and Jenkins put forward common step that sets up the model time series, which it is model identification, parameter estimation, model applicability test and prediction. While making use of this step to set up the model of time series, several model of well reflectedly actual data usually appear , carrying on the model choice is needed at this time. It is difficult to select model by using of traditional standard of model selection ,which only from a certain or two identity of time series to select model. In this paper the method based the traditional standard of model selection was put forward to select model from much identity of time series. Technique for order preference by similarity to Ideal solution consider selecting model from the mean value, standard deviation, skewness, kurtosis, adjusted R-squared, Akaike information criterion, schwarz criterion , sum squared resid, and the first order autocorrelation. And set up the model of rate of wangweigaoxin stock payoff to take into the the substantial evidence application.