部分线性模型的adaptive group lasso变量选择

Variable selection for partially linear models via adaptive group lasso

对部分线性模型的aglasso(adaptive group lasso)参数估计及变量选择问题进行研究.通过构造aglasso的估计函数,将分组部分线性模型变量的选择问题转化为分组因子的选择问题.理论研究表明:该方法能相合地识别真实模型,并且估计具有oracle性质.最后通过模拟研究了所提方法的有限样本性质.

T his paper study the aglasso （adaptive group lasso ） for simultaneous parameter estimation and variable selection to a partially linear model . The selection of group variable for partially linear model is translated into seclection of group factor . The theoretical results show that the new method is able to identify the true model consistently , and the resulting estimator has the oracle property . T he finite sample performance of the proposed procedures is illustrated by simulation studies .