The Application Of State Space Model In Processing Seasonal Time Series

Time series analysis, an important statistical analysis method, is widely used in natural science, social science, scientific research and human intelligence. It is a kind of important data, by which one can know the features of the structure of the system, its operational rule, and then to predict the future data and do control behavior. Due to the weather condition, customs and other reasons, many data show obvious seasonal cycle fluctuations. This thesis introduces a new proposed kind of method of time series analysis, namely State Space Model Method. The seasonal, tread, irregular component can be easily extracted easily via Kalman filtering method. The method also provides the reasonable structural framework for seasonal series.First, this paper discusses in details the Kalman filtering to develop State Space Model in time series analysis. Second, the paper gives the algorithm of Maximum Likelihood Estimation (MLE), Expectation Maximum (EM) to estimate all unknown parameters. Finally, the historical data of the retail sales of consumer goods are used to build the model and predict the future data. Moreover, the predict accuracy is better than that of the traditional Seasonal Auto-Regressive and Moving Average (SARIMA) method, Basic Structure Model (BSM). Furthermore, the development situation of the future consumer markets is analyzed.