Robust model-free feature screening via quantile correlation

被引：22

作者：

Ma, Xuejun ^{[1
]}

Zhang, Jingxiao ^{[1
]}

机构：

[1] Renmin Univ China, Sch Stat, Ctr Appl Stat, Beijing 100872, Peoples R China

来源：

JOURNAL OF MULTIVARIATE ANALYSIS | 2016年 / 143卷

关键词：

Quantile correlation; Ultrahigh-dimensionality; Sure screening; Robustness; VARYING COEFFICIENT MODELS; VARIABLE SELECTION; REGRESSION;

D O I：

10.1016/j.jmva.2015.10.010

中图分类号：

O21 [概率论与数理统计]; C8 [统计学];

学科分类号：

020208 ; 070103 ; 0714 ;

摘要：

In this paper, we propose a new sure independence screening procedure based on quantile correlation (QC-SIS). The method not only is robust against outliers, but also can discover the nonlinear relationship between independent variables and dependent variable. We establish the sure screening property under certain technical conditions. Simulation studies are conducted to assess the performances of QC-SIS, sure independent screening (SIS), sure independent ranking and screening (SIRS), robust rank correlation screening (RRCS) and distance correlation-sure independent screening (DC-SIS). Results have shown the effectiveness and the flexibility of the proposed method. We also illustrate the QC-SIS through an empirical example. (C) 2015 Elsevier Inc. All rights reserved.

引用

页码：472 / 480

页数：9