贝叶斯动态线性模型的组合预测  

Combinatorial Forecast of Bayesian Dynamic Linear Models

作　　者：蒋金凤[1] 刘福升[2] 张孝令[2] 

机构地区：[1]青岛大学师范学院数学系,青岛266071 [2]山东科技大学信息科学与工程学院,泰安271019

出　　处：《山东科技大学学报（自然科学版）》2002年第3期16-17,24,共3页Journal of Shandong University of Science and Technology(Natural Science)

摘　　要：组合预测就是将随机变量或向量的点预测进行组合 ,这些预测是根据几个模型所给出的。一般的方法是求这些点预测的算术平均值。本文给出的是根据模型的预测精度确定权的贝叶斯动态线性模型的组合预测 。Combinatorical forecast means combination of point forecasts from several random variables or vectors, which are gotten by several models. The general method is to compute their arithmetical means. This paper gives combinatorial forecast of Bayesian dynamic linear model in which its weight is decided according to their forecast precision. It's forecast precision is higher than that of arithetical means.

关 键 词：贝叶斯动态线性模型 组合预测 预测精度 随机变量 点预测 贝叶斯预测 算术平均值 

分 类 号：O212.8[理学—概率论与数理统计]