数学物理学报 ›› 1998, Vol. 18 ›› Issue (S1): 133-138.

• 论文 • 上一篇    下一篇

非线性模型中的拟似然估计和约束拟似然估计

林路   

  1. 邵阳师范高等专科学校数学系 邵阳 422000
  • 收稿日期:1996-10-14 修回日期:1997-04-16 出版日期:1998-12-26 发布日期:1998-12-26

Quasi Likelihood Estimation and Constrained Quasi Likelihood Estimation in Nonlinear Regression Model

Lin Lu   

  1. Shaoyang Twachers' College, Shaoyang 422000
  • Received:1996-10-14 Revised:1997-04-16 Online:1998-12-26 Published:1998-12-26

摘要: 该文证明了,在非线性回归模型中,若以均方误差或均方误差矩阵为标准,拟似然估计是正则广义拟似然估计类中的最优估计,并讨论了拟得分函数最优性与拟似然估计最优性的关系.为改进拟似然估计,该文提出了一种约束拟似然估计,并证明了约束拟似然估计比拟似然估计有较小的均方误差.

关键词: 非线性回归模型, 拟得分函数, 拟似然估计, 正则广义拟似然估计类, 约束拟似然估计

Abstract: In nonlinear regression model, the optimality of quasi likelihood estimatiom in theclass of normal generalized quasi likelihood estirnation is proved.The relation between the opti-mality of quasi score function and quasi likelihood estimation is discussed. To improve on quasilikelihood estimation,a constrained quasi likelihood estimation is presented. For this estimation,the superiority over quasi likelihood estimation is proved.

Key words: Nonlinear regression model, quasi score function, quasi likelihood estimation, class of normal generalized quasi likelihood estimation, constrained quasi likelihood estimation, mean squared error