Statistics and Its Interface

Volume 14 (2021)

Number 2

A non-marginal variable screening method for the varying coefficient Cox model

Pages: 197 – 209

DOI: https://dx.doi.org/10.4310/20-SII628

Authors

Lianqiang Qu (School of Mathematics and Statistics, Central China Normal University, Wuhan, China)

Liuquan Sun (Institute of Applied Mathematics, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing, China)

Abstract

The varying coefficient model has become a very popular statistical tool for describing the dynamic effects of covariates on the response. In this article, we develop a new variable screening method for the varying coefficient Cox model based on the kernel smoothing and group learning methods. The sure screening property is established for ultra-high-dimensional settings. In addition, an iterative groupwise hard-thresholding algorithm is developed to implement our method. Simulation studies are conducted to evaluate the finite sample performances of the proposed method. An application to an ovarian cancer dataset is provided.

Keywords

Cox model, kernel smoothing, non-marginal screening, ultra-high-dimensionality, varying coefficient

2010 Mathematics Subject Classification

62G08, 62N01

The full text of this article is unavailable through your IP address: 3.144.235.195

This research was partly supported by the National Natural Science Foundation of China (Grant Nos. 11771431 and 11690015), Key Laboratory of RCSDS, CAS (No. 2008DP173182), and by the Hubei Natural Science Foundation of China (Grant No. 2018CFB256).

Received 30 August 2019

Accepted 15 July 2020

Published 22 December 2020