关键词: 线性模型, 随机缺失, 经验似然, 复合分位数回归, 逆概率加权

Abstract:

We proposed a weighted composite quantile regression method based on empirical likelihood in linear model with some covariates missing at random, and proved the large sample properties of the proposed method under the missing at random mechanism. The results show that the proposed method is computationally simple and the estimation efficiency of the regression parameters is higher than that of the inverse probability weighted method.

Key words: linear model, missing at random, empirical likelihood, composite quantile regression, inverse probability weighting