if nobody knows the answer to my question on time-dependent ROC (see below), does any body know how TP and FP are related to classes and samples (orders) in non-time dependent ROC? Any reference? Thanks
--- On Tue, 11/3/09, carol white <wht_crl at yahoo.com> wrote:
From: carol white <wht_crl at yahoo.com> Subject: ROC To: r-help at stat.math.ethz.ch Date: Tuesday, November 3, 2009, 1:08 PM My question would be related more to time-dependent ROC (survivalROC package): I would like to know how true positives (TP) and false positives (FP) are related to the order of survival times and diagnostic markers. That is, having the a set of TP and FP, each of them is calculated from which diagnostic marker and survival times. Let's take mayo data set. How could I match the TP and FP generated from survivalROC function to cut.values (output) and Stimes (input)? Best, Carol ? ? ?