基于组合预测包容性检验的单项模型选择

Stepwise Selection Method of Single Model for Combination Forecast Based on Encompassing Test of Combination Models

组合预测模型实际上是单项预测模型的信息进行选择利用的过程。最优组合中单项预测模型的选择是十分重要的问题,将包容性检验应用于组合预测单项预测模型的选择;给出了基于组合模型包容性检验的单项模型的选择方法和步骤,该方法主要是利用增加一个单项预测方法的组合预测模型与原组合预测模型之间的包容性检验,确定单项模型是否要保留在组合预测模型中,这样可以达到提高组合预测的预测有效度。最后通过应用实例的分析,表明通过组合预测模型的包容性检验筛选出合适的单项预测模型,再建立组合预测模型就能够达到提高预测精度的效果,因此该方法是可行的和有效的。

Combination forecasting model is actually the use of the individual prediction models. How to select the single models in combination forecast is very important. The encompassing test method of combination forecasting models is applied to select single models. The single model group selection method is presented based on encompassing test of combination forecasting models to im prove precision of combination forecasting model. The method is mainly to determine whether a single model to be retained in the combined forecasting model according to encompassing test between combined forecasting models of a single forecast method in- creased with the original combination forecasting model. Finally, the analysis of the application example shows that the predictive accuracy effect of combination forecasting model is improved according to the combined forecasting model encompassing test select single models, so that this method is feasible and effective.