预测精度和预测风险是预测学研究的核心问题.从信息互补的角度,组合预测提供了一条有效的途径.本文综述了组合预测的模型与方法的分类、权重计算方法、模型构建方法,并对最优子模型的选择等问题进行综述和探讨,最后展望了在不确定环境下组合预测方法和未来的研究方向.
Prediction accuracy and prediction risk are the core issues of prediction research. From the perspective of complementary information, combination forecasting provides an effective way to improve prediction accuracy. We give a survey of researches on the development of combination forecasting models and methodologies in this paper. We conduct the reviews from several aspects, which are the classification of combination forecasting models, the methods of weight calculation, some kinds of methods of model constructing for combination forecasting. The problems of the optimal sub-models selection in the combination forecasting are reviewed and discussed, and finally we give the prospects and future research directions of combination forecast models under uncertain environment.