报告人：朱仲义 教授

【Abstract】We propose a two-step variable selection procedure for high dimensional fixed censored quantile regression, in which the response subject to fixed censoring, and the dimension of the covariates, is very large, possibly much larger than the sample size . To account for fixed censoring data with multivariate covariates, we employ the ideas of informative subset-based estimator (Tang et el. (2012)) and effective dimension reduction. Under some regularity conditions, we show our procedure enjoys the model selection consistency, as long as the conditional censoring probability can be estimated consistently. We demonstrate that the first step penalized estimator with LASSO penalty can reduces the model from ultra-high dimensional to a model whose size has the same order with the true model, and the selected model can cover the true model. The second step excludes the remained irrelevant covariates by applying adaptive LASSO penalty to the reduced model obtained from the first step. We conduct a simulation study and a real data analysis to evaluate the finite sample performance of the proposed approach.

时间：2013年10月15日（周二）下午15:45开始

地点：数学系致远楼107室