摘要
预测是统计学实际应用的一个主要方面,多元线性回归预测是一种很好的方法,广泛地应用在各种实际领域,但其局限性及不足也是明显的。本文以一种新的观点认识数据,即认为变量的观测里均含有误差,同时认为不应删除经慎重选择进来的解释变量。为此,本文提出了一种新的多元预测方法———多元线性EIV预测。本文还考虑了新预测模型的一个实例应用,并从相对偏差上与多元回归预测进行了比较,从而揭示了多元线性EIV预测的先进性及较好的预测精度。
Prediction is a main aspect of statistical practical application.Though the multivariate linear regression prediction is one of the very good methods and is applied widely to various kinds of actual fields,but its limitation and insufficient is obvious.This paper understands data by one new viewpoint that the observation of all variables contains the error,and thinks that the explanation variable chosen cautiously needn’t delete,For this reason,this paper put forward a kind of new multivariate prediction method—a multivariate linear EIV prediction. Bt considering an illustration of this new prediction model and comparing the relative deviation between EIV prediction model and multivariate regression prediction model, this paper shows that the multivariate linear EIV prediction has better precision of prediction.
出处
《数理统计与管理》
CSSCI
北大核心
2005年第2期55-59,共5页
Journal of Applied Statistics and Management
基金
教育部留学回国人员启动基金及南京理工大学科研发展基金资助
