主讲人：Xingqiu Zhao, The Hong Kong Polytechnic University

时  间：2014年12月4日（周四）  10：00-11：00

地  点：必赢76net线路官网北一区文科楼707教室

摘  要：The partially linear  Cox model assumes that it is known a priori which covariates have a linear effect and which do not on the log hazards function.  However, this is rarely known in practice. We propose a semiparametric pursuit method to simultaneously detect and estimate  linear and nonlinear covariate effects on the log hazards function through a penalized group selection method with concave folded penalties. The unknown smoothing functions of the nonlinear component are approximated by the B-splines. Both the parametric and nonparametric estimators are consistent, and the parametric estimator is asymptotically normal. We develop a modified blockwise majorization descent algorithm that is easy to implement and has a fast convergence rate. Simulation studies indicate that the proposed method works well, and the primary biliary cirrhosis data are analyzed for illustration.