吉林大学学报(理学版)

• 数学 • 上一篇    下一篇

协变量缺失下变系数模型基于经验似然的加权分位数回归

袁晓惠, 鞠婷婷   

  1. 长春工业大学 基础科学学院, 长春 130012
  • 收稿日期:2016-03-02 出版日期:2017-03-26 发布日期:2017-03-24
  • 通讯作者: 袁晓惠 E-mail:yuanxh@ccut.edu.cn

Weighted Quantile Regression for VaryingCoefficient Modelswith Missing Covariates Based on Empirical Likelihood

YUAN Xiaohui, JU Tingting   

  1. School of Basic Science, Changchun University of Technology, Changchun 130012, China
  • Received:2016-03-02 Online:2017-03-26 Published:2017-03-24
  • Contact: YUAN Xiaohui E-mail:yuanxh@ccut.edu.cn

摘要: 在部分协变量随机缺失的变系数分位数回归模型中, 提出回归参数的逆概率加权估计和基于经验似然的加权估计, 并讨论了这两种估计的大样本性质. 从渐近方差可见, 基于经验似然的加权估计效率高于逆概率加权估计.

关键词: 分位数回归, 经验似然, 协变量缺失, 逆概率加权估计, 变系数

Abstract: We proposed the inverse probability weighted (IPW) estimation of regression parameters and the empirical likelihoodbased weighted (ELW) estimation for varyingcoefficient quantile regression model with partial missing covariates, and discussed the large sample properties of these two estimations. The ELW estimation is more efficient than the IPW estimation by the asymptotic variance.

Key words: quantile regression, empirical likelihood, missing covariates, inverse probability weighted (IPW) estimation, varying-coefficient

中图分类号: 

  • O212.4