基于向量自回归和多元线性回归的区间序列预测方法  

A prediction method of interval series based on vector autoregression and multiple linear regression

作　　者：乌兰 曾祥艳[1] 周伟 WU Lan;ZENG Xiangyan;ZHOU Wei(School of Mathematics and Computational Science,Guilin University of Electronic Technology,Guilin 541004,China)

机构地区：[1]桂林电子科技大学数学与计算科学学院,广西桂林541004

出　　处：《桂林电子科技大学学报》2021年第3期253-258,共6页Journal of Guilin University of Electronic Technology

基　　金：国家自然科学基金(71801060);国家级大学生创新创业训练计划(202010595043)。

摘　　要：为了使自回归模型和多元线性回归模型能直接对区间时间序列建模,将区间看作一个二维列向量,将2个模型的参数设置为矩阵型,提出了面向区间序列的向量自回归模型、向量多元线性回归模型以及两者的组合模型。分析表明,新模型实质上是结合了区间的上下界点序列对其中一个界点进行预测,体现了区间的整体性和上下界点之间的内在联系。将该模型对发电量和客运量进行区间预测,结果表明,模型的预测效果良好。In order to make the autoregression model and the multiple linear regression model can be directly built on the interval time series modeling,the interval is regarded as a two-dimensional column vector and the parameters of the two models are set as the matrix.The vector autoregressive model,the vector multiple linear regression model and their combination model are presented.The analysis shows that the new model essentially combines the sequence of upper and lower boundary points to predict one of them and embodies the integrality of interval and the internal relation between upper and lower boundary points.The proposed model is used to predict the power generation and the passenger traffic,and the results show that the prediction effect is good.

关 键 词：向量自回归 向量多元线性回归 区间序列 组合模型 

分 类 号：O213[理学—概率论与数理统计] N32[理学—数学]