A Combination Forecasting Model Based on Time Series Variable Weighting Coefficient
Guochao Yang · 2012
To improve the forecasting model's accuracy,this paper presented an optimized combination forecasting model based on analysis of time series variable weighting coefficient.With the Least Sum of Square Error as the standard,this study established a combination forecasting model whose variable weighting coefficient is nonnegative,and calculated the optimized weighting coefficient of individual forecasting model in different time.Using the ARMA model,the study predicted the variable weighting coefficients of individual forecasting models and figured out the combination forecasting model.The example showed that this combination forecasting model is better than individuals.As a result,it can make the best of the information of individuals and lead to obtaining least Sum of Square Error of the model.Therefore,the model is more accurate and applicable.