Sufficient variable selection using independence measures for continuous response

JOURNAL OF MULTIVARIATE ANALYSIS

卷:173

Article

Yang, Baoying¹  Yin, Xiangrong²  Zhang, Nan³ 

[1] Southwest Jiaotong Univ, Coll Math, Dept Stat, Chengdu 611756, Sichuan, Peoples R China

[2] Univ Kentucky, Dept Stat, 319 Multidisciplinary Sci Bldg, Lexington, KY 40536 USA

[3] Univ Georgia, Dept Stat, 204 Stat Bldg,101 Cedar St, Athens, GA 30602 USA

关键词: Distance correlation;    Hilbert-Schmidt independence criterion;    Independence test;    Marginal feature screening;    Sufficient variable selection;   

DOI  :  10.1016/j.jmva.2019.04.006

来源: Elsevier

【 摘 要 】

We propose two sufficient variable selection procedures, i.e., one- and two-stage approaches using independence measures for continuous response, illustrated by distance correlation and the Hilbert-Schmidt Independence Criterion correlation. We show the advantages of the proposed procedures over some existing marginal screening methods through simulations and a real data analysis. Our procedures are model-free and thus robust against model mis-specification. They are particularly useful when some active predictors are marginally independent of the response. (C) 2019 Elsevier Inc. All rights reserved.

【 授权许可】

Free