Computational Statistics

, Volume 28, Issue 4, pp 1835–1852

Sparse dimension reduction for survival data

Original Paper

DOI: 10.1007/s00180-012-0383-4

Cite this article as:
Yan, C. & Zhang, D. Comput Stat (2013) 28: 1835. doi:10.1007/s00180-012-0383-4


In this paper, we study the estimation and variable selection of the sufficient dimension reduction space for survival data via a new combination of \(L_1\) penalty and the refined outer product of gradient method (rOPG; Xia et al. in J R Stat Soc Ser B 64:363–410, 2002), called SH-OPG hereafter. SH-OPG can exhaustively estimate the central subspace and select the informative covariates simultaneously; Meanwhile, the estimated directions remain orthogonal automatically after dropping noninformative regressors. The efficiency of SH-OPG is verified through extensive simulation studies and real data analysis.


Censored dataHazard functionVariable selection Dimension reduction

Copyright information

© Springer-Verlag Berlin Heidelberg 2012

Authors and Affiliations

  1. 1.Department of Finance and Insurance, School of Economics, Center of Research of Finance Econometrics and Risk ManagementNanjing UniversityNanjingChina