Nonparametric estimation in the Cox model (Q2366730)

Using multivariate counting processes the paper derives nonparametric estimators of the relative risk in a generalized Cox regression model with covariates. The estimators are derived using the penalized partial likelihood approach. The paper considers the asymptotic convergence of these estimators in the Sobolev \(W^{bm}_ 2\) norm and obtains upper bounds on the rate of convergence. These upper bounds match the optimal rates available for linear nonparametric regression and estimation.