统计与管理学院2015年学术报告第10期
【主 题】 Imputation Methods and Efficient Estimation for Linear Models with Missing Responses
【报告人】 唐年胜 教授
云南大学
【时 间】 2015年3月23日（星期一）10:30－11:30
【地 点】 上海财经大学统计与管理学院大楼1114室
【语 言】 英文
【摘 要】 The aim of this paper is to propose nonparametric and parametric imputation approaches for parameter estimation in a quantile regression model when some responses are missing at random. The two imputation approaches can successfully remove selection bias due to missing data, while the proposed parametric imputation approach yields more reliable results even in high dimensional problems. To increase statistical efficiency, we propose a weighted composite quantile estimation and a weighted quantile average estimation. Moreover, for the parametric imputation approach, we use the SCAD and adaptive-LASSO penalties to simultaneously select significant variables and estimate unknown parameters. We establish the asymptotic properties of the proposed estimators under both imputation methods. An efficient algorithm is developed for fast implementation of the proposed methodologies. We also discuss a model selection criterion, which is based on the so-called IC_Q statistic, to select the penalty parameters. The performance of the proposed methods is illustrated via simulated and real data sets.
