| It is of great importance to establish a combined forecasting model which is in line with the reality of economic forecasting.Since Bates and Granger(1969)put forward the combined forecasting method in the late 1960s,combined forecasting has become one of the hot issues in economics and management.However,traditional combined forecasting models are usually established based on some specific criteria,ignoring the characteristics and the actual needs of economic forecasting:Most combined forecasting models only take one single criterion as the optimization objective,and fail to combine different criteria at the same time.On the other hand,the forecasting effect of each single forecasting model often behaves quite differently.Therefore,it is of great theoretical and practical significance to establish a pertinent combined forecasting model which accords with the reality of economic forecasting.In economic forecasting,it is often necessary to control the maximum forecasting error reasonably while improving the forecasting accuracy.Therefore,based on the existing literature,by combining the sum of squared error and maximum absolute error,this thesis first establishes an improved variable weight combined forecasting model based on mixed index,some basic properties and the solution method of this model are studied as well.Secondly,based on the above-mentioned improved variable weight combined forecasting model,by using the total energy consumption data of China from 1978 to 2017,the combined forecasting results of China's total energy consumption are obtained.The empirical results show that:(1)This model can effectively reduce the forecasting error,disperse the forecasting risk of the single forecasting method;(2)The combined forecasting model also has high out-of-sample forecasting accuracy.In conclusion,from the empirical point of view,this study verifies the effectiveness of the combined forecasting model based on mixed index. |
PDF Full Text Request