摘要
传统的组合预测模型中每一种单项预测方法在各个时点具有相同的加权系数,但实际上同一种单项预测方法在各个时点的预测精度有高有低,为了克服单项预测方法取固定权系数的缺陷,构建了基于一种贴近度的IOWA算子的变权系数的组合预测模型,并探讨模型的非劣性组合预测、优性组合预测存在性的充分条件,实例分析结果表明:该模型在预测效果评价指标体系中明显优于传统的组合预测方法。
In the traditional combination forecasting model each single forecast method has me same weighting factor at each time. Actually to the same individual forecasting method, it has high and low prediction accuracy at each time. In order to overcome the single forecasting method to take the fixed weight coefficient of defects, the variable weight coefficients of combination forecast model is constructed based on one kind of approach degree and induced ordered weighted averaging （IOWA） operator, and the existence of no inferior combination forecasting and superior combination forecasting are discussed in this paper. Finally, an example shows that the new model in predicting the effects of evaluation index system is obviously superior to the traditional combination forecasting method.
出处
《统计与信息论坛》
CSSCI
2010年第2期32-37,共6页
Journal of Statistics and Information
基金
安徽省教育厅人文社会科学研究资助项目<组合预测方法及其在安徽经济发展预测中的应用研究>(2009sk130)
关键词
组合预测
贴近度
诱导有序加权平均算子
优性组合预测
combination forecasting
closeness degree
IOWA operator
superior combination forecasting
