统计与数学学院2016手拉手Workshop系列之（二）：Advanced regression models and inferences
发布日期：2016-05-05主题：Advanced regression models and inferences
时间：2016年5月8（星期日）9：00-17：00
地点：沙河校区，4号楼107
具体日程安排如下：
09：00-10：30 短期课程： Variable selection in high-dimensional Tobit models
主讲人：Professor Hua Liang, Department of Statistics, George Washington University
摘要：One of the most challenging tasks in regression analysis is choosing which variables to include in the regression equation. If an important variable is omitted from a regression, we simply miss the effect of this important variable. The coefficient estimates become biased. Associated forecasts are also going to be biased. If an unimportant variable is included in a regression, the standard errors of the coefficient estimates get larger. This will also increase the standard error in forecasts. The larger standard errors are going to generate smaller test statistics and variable that may be important may be found insignificant by hypothesis tests. These concerns become more serious in high-dimensional settings. The goal of this short course is to introduce penalization regression and penalization based variable selection for high-dimensional linear models with an emphasis on an extension to the Tobit model. The short course will focus on the ideas of these innovative approaches, a comparison with the well-established methods. Real data will be provided for illustration during the short course.
10：30-11：30 张新雨中国科学院数学与系统科学研究院
Title: Inference after Model Averaging in Linear Regression Models
Abstract: This paper considers the problem of inference for nested least squares averaging estimator. We study the asymptotic behavior of the Mallows model averaging estimator (MMA; Hansen, 2007) and the jackknife model averaging estimator (JMA; Hansen and Racine, 2012) under the standard asymptotics with fixed parameters setup. We find that both MMA and JMA estimators asymptotically assign the zero weight to the under-fitted model, and MMA and JMA weights of just-fitted and over-fitted models are asymptotically random. Building on the asymptotic behavior of model weights, we derive the asymptotic distributions of MMA and JMA estimators and propose a valid confidence interval for the least squares averaging estimator. Monte Carlo simulations show that the coverage probability of proposed confidence intervals achieves the nominal level. We apply our method to examine the effect of the student-teacher ratio on student achievement. (Joint work with Chu-An Liu)
11：30-12：00 讨论
14：00-14：45 刘玉涛：Variable selection in nonparametric Tobit models
15：00-15：45 潘蕊：An approximate least squares methods for spatial autoregression with covariates
16：00-17：00 讨论
[编辑]：孙颖
