摘要
研究了线性回归中方差加权和均值漂移混合模型中LSE和若干有偏估计的局部影响分析.从基于回归系数无偏估计的似然函数出发,求出了回归系数最小二乘估计(LSE)在混合扰动下的影响曲率、最大影响曲率和最大曲率方向,对于有偏估计、在后验分布对数似然函数的基础上,求出了这些有偏估计在混合扰动下的影响曲率、最大影响曲率和最大曲率方向.
This paper studies the 1ocal influence analysis on mixed linear regression model with case-weights and mean-shift simultaneously. From likelihood function (based on unbiased estimate of regression coefficients), the influence curvature and maxima1 influence curvature and maximal influence curvature direction for LSE of regression coefficients uirder the mixed model are derived and from a posterior distribution logarithmic Iikelihood function the influence curvature and maximal influence curvature and maximal influence curvature direction for some biased estimates of regression coefficients under the mixed model are derived.
出处
《广西师范大学学报(自然科学版)》
CAS
1997年第4期25-31,共7页
Journal of Guangxi Normal University:Natural Science Edition
关键词
岭估计
广义岭估计
混合扰动模型
有偏估计
mixed model
ridge estimate
generalized ridge estimate
influence curvature