删失回归模型中SCAD型变量选择与估计(英文)  

Variable selection and estimation via SCAD-type penalty for censored regression models

作　　者：刘显慧[1] 王占锋[1] 吴耀华[1] 

机构地区：[1]中国科学技术大学统计与金融系,安徽合肥230026

出　　处：《中国科学技术大学学报》2013年第3期182-189,共8页JUSTC

基　　金：Supported by Natural Science Foundation of China(10871188,11101396)

摘　　要：删失回归模型是一种很重要的响应变量受限制的模型,它广泛应用于计量经济学中.基于SCAD罚函数,提出了SCAD型罚变量选择方法.该方法可选出重要的回归变量,即真参数中非零系数,同时给出非零参数相合估计.另外,证明了变量选择方法是相合的,具有oracle性质.最后,进行数值模拟计算说明所提出方法的效果。Censored regression （＂Tobit＂） model is a kind of limited dependent variable model, which is widely used in econometric research. Based on the smoothly clipped absolute deviation （SCAD）, for censored regression model, a SCAD-type variable selection method was developed, which selects the non-zero coefficients corresponding to the significant variables and simultaneously gives a consistent estimate of parameter. The variable selection procedure is consistent and possesses the oracle properties. Finally, numerical studies were conducted to evaluate the performance of the proposed method for censored regression models.

关 键 词：删失回归模型 变量选择 oracle性质 

分 类 号：O212.1[理学—概率论与数理统计]