数学物理学报 ›› 1997, Vol. 17 ›› Issue (1): 38-46.

• 论文 • 上一篇    下一篇

线性相关模型中误差方差的经验似然估计及其Bootstrap

石坚   

  1. 中国科学院系统科学研究所 北京 100080
  • 收稿日期:1995-01-16 修回日期:1995-09-22 出版日期:1997-02-26 发布日期:1997-02-26

An Empirical Likelihood Estimation of the Error Variance in Linear Correlation Models and its Bootstrap

Shi Jian   

  1. Institute of Systems Science, Academia Sinica, Beijing 100080
  • Received:1995-01-16 Revised:1995-09-22 Online:1997-02-26 Published:1997-02-26

摘要: 该文利用经验似然方法,对线性相关模型中误差方差的传统最小二乘型估计进行修正,得到的修正估计其渐近方差比传统估计的更小.同时,我们还讨论了修正估计的Bootstrap逼近问题.

关键词: 相关模型, 误差方差, 最小二乘, 经验似然, Bootstrap

Abstract: In this paper,we use the empirical likelihood method to modify the least-square type estimation of the error variance in linear correlation model. It is shown that the modified estimation has smaller asymptotic variance than the traditional one. Meanwhile,we have given a bootstrap approximation for the modified estimation.

Key words: Correlation model, error variance, least square, empirical likelihood, bootstrap