基于密度算子的组合预测方法  

Combination Forecasting Method Based on Density Operator

作　　者：易平涛[1] 郭亚军[1] 李伟伟[1] 

机构地区：[1]东北大学工商管理学院,辽宁沈阳110004

出　　处：《预测》2011年第4期60-64,共5页Forecasting

基　　金：国家自然科学基金资助项目(71071030;71071031);中国博士后科学基金特别资助项目(200902545);辽宁省博士启动基金资助项目(20081020)

摘　　要：考虑到单项预测方法在不同时刻预测准确性的差异,提出了一种基于密度算子的组合预测方法。该方法的核心思想是利用样本区间内单项预测方法的预测准确性对预测值聚类;然后基于密度算子的思想对单项预测法集结,得到一种新的组合预测模型;在此基础上,基于预测值和相应观察值误差最小的思想确定单项预测法的加权系数。该模型充分利用了样本区间内单项预测法的预测准确性,因而预测值具有较高的准确性。最后,给出一个实例,并对其结果进行比较分析,证明了该组合预测模型的有效性。Considering the difference of single forecasting method＇ s veracity at different times, this paper presents a new combination forecasting model based on density operator. The core ideal of this method is to cluster the forecasting values by the forecasting veracity, then build the combination forecasting model using the density operator. Based on this, the single forecasting method weight is determined through the minimal warp between the forecasting values and the real val- ues. This combination forecasting model makes full use of single forecasting model＇ s veracity, so the forecasting value veracity has been increased. At last, an application example is given, and its result is analyzed to illustrate the validity of the new combination forecasting model.

关 键 词：组合预测 密度算子 预测准确度 密度加权向量 

分 类 号：C934[经济管理—管理学]