IJPAM: Volume 22, No. 4 (2005)

INFLUENCE ANALYSIS OF BLUE IN
RESTRICTED LINEAR REGRESSION MODELS

Shangli Zhang
School of Science
Beijing Jiaotong University
Beijing, 100044, P.R. CHINA
e-mail: shlzhang@center.njtu.edu.cn


Abstract.This paper is devoted to discussing the influence problem of Best Linear Unbiased Estimate (BLUE) in restricted linear regression models. The influence on BLUE with covariance matrix disturbing or data missing is studied. The relationships between $\hat{\beta}_{L}(G)$ and $\hat{\beta}_{L}$ are established, where $\hat{\beta}_{L}(G)$ is the BLUE of $\beta$ in mode ([*]) and $\hat{\beta}_{L}$ is the BLUE of $\beta$ in model ([*]). We define the generalized Cook distance $D_{G}$ to assess the disturbance influence and obtain a formula for $D_{G}$.

Received: May 28, 2005

AMS Subject Classification: 62J20

Key Words and Phrases: linear regression model, best linear unbiased estimate, influence analysis

Source: International Journal of Pure and Applied Mathematics
ISSN: 1311-8080
Year: 2005
Volume: 22
Issue: 4