Variable selection procedure from multiple testing
Variable selection procedure from multiple testing
复制标题
基于潜在低维结构的空间和时间克里金法
DOI：
10.1007/s11425-000-0000-0
复制
发表时间：
2018
期刊：
影响因子：
--
通讯作者：
Shurong Zheng
中科院分区：
文献类型：
--
作者：
Baoxue Zhang;Guanghui Cheng;Chunming Zhang;Shurong Zheng
在过去的十年中，变量选择在统计学习和科学发现中扮演着重要的角色，多重检验是统计推断中的一个基本问题，在许多科学领域中有着广泛的应用。在这两个领域分别取得了重大进展。本文的目的是找出自适应套索与线性回归模型中多重检验方法之间的联系，并提出基于多重检验方法的变量选择方法和控制选择错误率的方法。仿真研究表明，所提出的方法在控制选择错误率和实现大功率在很宽的设置范围内的良好性能。
Variable selection plays an important role in statistical learning and scientific discoveries during the past ten years and multiple testing is a fundamental problem in statistical inference, also with wide applications in many scientific fields. Significant advances have been achieved in both two areas, respectively. This paper aims at figuring out a connection between the adaptive lasso and multiple testing procedure in linear regression models, and also aims at proposing procedures based on the multiple testing methods to select variables and control the selecting error rate which is called false discovery rate. Simulation studies show good performance of the proposed methods on controlling the selecting error rate and achieving great powers in a wide range of settings.