摘要
在局部影响分析中,人们往往对线性模型的单个回归参数感兴趣.根据Wu&Luo(1993)的二阶局部影响分析方法,在这篇文章里我们通过对参数的MLE扰动曲面的研究来检验线性模型中观测值对单个回归参数的局部影响.一个实际例子说明了该方法在识别强影响点时的有效性.
An individual coefficient in a linear regression model are often of special interest in the analysis of local influence.Multivated by Wu & Luo(1993)'s second order approach to local influence,in this paper we assess the local influence of the observations on a selected coefficient in linear regression moldel by studying the surface formed by the perturbation and the MLE of the selected coefficient.An example demonstrates the effectiveness of this method to the identification of influential observations on model fitting.
出处
《南开大学学报(自然科学版)》
CAS
CSCD
1996年第2期28-33,共6页
Acta Scientiarum Naturalium Universitatis Nankaiensis
关键词
二阶方法
局部影响
线性模型
回归参数
a selected regression coefficient
second order approach
deletion diagnostic
local innuence
influential case