Premium
Kernel estimates of hazard functions for carcinoma data sets
Author(s) -
Horová Ivana,
Zelinka Jiří,
Budíková Marie
Publication year - 2006
Publication title -
environmetrics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.68
H-Index - 58
eISSN - 1099-095X
pISSN - 1180-4009
DOI - 10.1002/env.743
Subject(s) - kernel (algebra) , hazard , kernel density estimation , bandwidth (computing) , kernel method , mathematics , function (biology) , computer science , statistics , econometrics , artificial intelligence , discrete mathematics , support vector machine , estimator , telecommunications , chemistry , organic chemistry , evolutionary biology , biology
The present article focuses on kernel estimates of hazard functions and their derivatives. Our approach is based on the model introduced by Müller and Wang (1990). In order to estimate the hazard function in an effective manner an automatic procedure in a paper by Horová et al. (2002) is applied. The procedure chooses a bandwidth, a kernel and an order of a kernel. As a by‐product we propose a special procedure for the estimation of the optimal bandwidth. This is applied to the carcinoma data sets kindly provided by the Masaryk Memorial Cancer Institute in Brno. Attention is also paid to the points of the most rapid change of the hazard function. Copyright © 2006 John Wiley & Sons, Ltd.