A General Class of Estimators for the Linear Regression Model Affected by Collinearity and Outliers

被引：1

作者：

Macedo, Pedro ^{[1
]}

Scotto, Manuel ^{[1
]}

Silva, Elvira ^{[2
]}

机构：

[1] Univ Aveiro, Dept Matemat, P-3810193 Aveiro, Portugal

[2] Univ Porto, Fac Econ, P-4100 Oporto, Portugal

来源：

COMMUNICATIONS IN STATISTICS-SIMULATION AND COMPUTATION | 2010年 / 39卷 / 05期

关键词：

Collinearity; Linear regression; Maximum entropy; Outliers; Robust regression; MAXIMUM-ENTROPY APPROACH; STATISTICAL MECHANICS; INFORMATION THEORY; RIDGE REGRESSION; ECONOMETRICS;

D O I：

10.1080/03610911003695719

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

In this article, a general class of estimators for the linear regression model affected by outliers and collinearity is introduced and studied in some detail. This class of estimators combines the theory of light, maximum entropy, and robust regression techniques. Our theoretical findings are illustrated through a Monte Carlo simulation study.

引用

页码：981 / 993

页数：13

共 42 条

[1]

[Anonymous], 2006, QED: The strange theory of light and matter

[2]

[Anonymous], TECHNOMETRICS

[3]

[Anonymous], 2004, ROBUST STAT

[4] Efficient frontier estimation: a maximum entropy approach [J].