缺失数据下分位数回归模型的平均估计

Model Averaging in Quantile Regression with Missing Data

本文主要探究了缺失数据下分位数回归模型的模型平均问题。首先基于协变量平衡倾向得分方法构造出候选模型中回归参数的加权分位数回归估计,并推导出其渐近分布;接着构造兴趣参数的模型平均估计量,并给出其大样本性质;最后构造出一个覆盖概率趋于预期水平的置信区间。

This paper mainly explores the model averaging problem of quantile regression models with missing data. Firstly, a weighted quantile regression estimator of the regression parameters in the candidate model is constructed based on the covariate balancing propensity score method, and then its asymptotic distribution is derived. Secondly, the model averaging estimator of the focus parameter is constructed, and its large sample property is obtained. Finally, a confidence interval is constructed with a coverage probability that tends to the expected level.