Research On The Application Of Combined Model In Forecasting The Total Retail Sales Of Social Consumer Goods In China

The total retail sales of social consumer goods is not only an important indicator to measure the level of consumption of our people,but also an important indicator of the national economy.Therefore,the analysis of the development trend of China's total retail sales of social consumer goods is of great significance for the healthy development of China's economy.The total retail sales of social consumer goods is a set of time series,according to the classic time series forecasting theory,the specific work of this paper includes:Firstly,to build a X-12-ARIMA model(addition and multiplication)and compare these two models,the results show that the fitting effect of X-12-ARIMA model is higher than that of X-12-ARIMA addition multiplication model,The MAPE of the X-12-ARIMA multiplication model is smaller and the degree of fitting is higher.Compared with the traditional time series forecasting methods AR?MA and ARIMA,the results show that the fitting degree of X-12-ARIMA multiplication model is smaller,and it has certain advantages to forecast.Secondly,the exponential smoothing method(ETS method)under the state space model is constructed.The exponential smoothing theory under the state space model is studied systematically,and the point prediction method of the commonly used exponential smoothing method is given.Through the empirical analysis,the results show that the fitting degree and prediction accuracy of the ETS model are relatively high,and the MAPE of the model is smaller.The model is better for the seasonal,trend and periodic factors of the original time series.The ETS model can adequately analyze and eliminate the information contained in the original timing.Thirdly,according to the fitting effect of the single prediction model,this paper constructs the combination forecasting model.On the basis of this,two kinds of optimal weight coefficient algorithms are introduced,which are Linear programming method and Chaos Particle Swarm Optimization.According to the empirical analysis,the results show that the combinatorial model based on chaotic particle swarm optimization has higher degree of fitting,and the fitting effect is higher than the single prediction method.The chaotic particle swarm optimization algorithm is used to optimize the weight coefficient,and the fitting precision and prediction precision of the model are improved to a great extent.The combined model fully extracts the advantages of each individual model,and combines the advantages of the single prediction model together.Finally,according to the results of the above single model and the combined model,the results show that the combination model based on the chaotic particle swarm optimization algorithm to optimize the weight coefficient is high,and the forecast of the total retail sales of social consumer goods is better,and the error is smaller.And the combination of the two types of weight optimization methods established in this paper is used to compare and analyzethe time series of the total retail sales of social consumer goods in China.Meanwhile,the forecast of the total retail sales of social consumer goods in China is forecasted.In summary,the fitting precision of the fitting accuracy combination model was higher than that of the single forecasting model,and in the composite structure,using chaotic particle swarm optimization can further improve the fitting precision of the combined model combined weight coefficient,so the combination forecasting model established in this paper is effective and has certain practical value and guiding significance.