基于成分数据的变权重组合预测的权重确定

发布时间：2018-04-07 15:05

本文选题：组合预测　切入点：变权重　出处：《山西大学学报(自然科学版)》2014年02期

【摘要】：变权重组合预测相比单预测、定权重组合预测能提高预测精度,其关键是确定权重。变权重组合预测模型权重的确定有很多方法,而预测期权重的确定很少受人关注。已有的确定预测期权重的方法是分别对每种预测方法的权重作预测,再归一化处理。文章基于成分数据的知识来确定预测期的权重,与已有方法相比,该方法不仅能保证预测期的权重求和为1,而且能降低维数。实验分析表明:分别模拟50,100,150,…,1 000次该方法预测精度比已有方法高,所占的比例大于50%;实例分析中该方法的MSE小于已有方法的MSE。从实验分析和实例分析可以看出,该方法能提高预测精度,是一种行之有效的方法。
[Abstract]:Compared with single prediction, variable weight combination prediction can improve the accuracy of prediction, and the key is to determine the weight.There are many methods to determine the weight of variable weight combination forecasting model, but few people pay attention to determining the weight of forecasting option.The existing methods to determine the weight of forecasting options are to predict and normalize the weights of each forecasting method respectively.In this paper, the weight of prediction period is determined based on the knowledge of component data. Compared with the existing methods, this method can not only ensure that the sum of the weights of the prediction period is 1, but also reduce the dimension.The experimental analysis shows that: the simulation 50100150, respectively...The prediction accuracy of this method is higher than that of the existing method, and the proportion of this method is more than 50 times, and the MSE of this method is smaller than that of the existing method in the case analysis.From the analysis of experiments and examples, it can be seen that this method can improve the accuracy of prediction and is an effective method.
【作者单位】：山西大学数学科学学院;计算智能与中文信息处理教育部重点实验室;
【基金】：国家自然科学基金重点项目(71031006) 国家自然科学青年基金(41101440) 山西省研究生教育改革研究课题(20102003)
【分类号】：O212.1

【参考文献】