Two-stage model selection procedures in partially linear regression
Two-stage model selection procedures in partially linear regression
复制标题
DOI：
10.2307/3315936
复制
发表时间：
2004-06-01
影响因子：
0.6
通讯作者：
Wegkamp, MH
中科院分区：
文献类型：
--
作者：
Bunea, F;Wegkamp, MH
对部分线性模型Y = f（0）（T）+X 'beta（0）+X' beta（0）提出了一个两阶段估计方法.它们展示了如何一致地估计beta（0）的非零分量的位置。在惩罚最小二乘的情况下，他们的方法与Besov球上f（0）的Minimax自适应估计是兼容的。他们的证明是基于一种新的预言不等式。
The authors propose a two-stage estimation procedure for the partially linear model Y = f(0) (T) + X'beta(0) + epsilon. They show how to estimate consistently the location of the nonzero components of beta(0). Their approach turns out to be compatible with minimax adaptive estimation of f(0) over Besov balls in the case of penalized least squares. Their proofs are based on a new type of oracle inequality.