Hi everybody. I'm trying to fit a weibull survival model with a spline
basis for the predictor, using the survival library. I've noticed that it
doesn't seem to be possible to use the aic method to choose the degrees of
freedom for the spline basis in a parametric regression (although it's
fine with the cox model, or if the degrees of freedom are specified directly
by the user), and I was wondering if there is some reason for this?
Simple example:
library(survival)
attach(cancer)
# cox model with specified degs of freedom - works fine
fit1 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,3), cancer)
# cox model with aic - works fine
fit1 <- coxph(Surv(time, status) ~ ph.ecog + pspline(age,method="aic"),
cancer)
# weibull model with specified degs of freedom - works fine
fit1 <- survreg(Surv(time, status) ~ ph.ecog + pspline(age,3), cancer,
dist="weibull")
# weibull model with aic - gives an error
fit1 <- survreg(Surv(time, status) ~ ph.ecog +
pspline(age,method="aic"),dist="weibull", cancer)
Error in (cfun[[1]])(parmlist[[1]], iter, iterlist[[1]], n.eff, pdf[1], :
object 'loglik' not found
I also get a similar error about "object 'loglik' not found" if I try to use
method="aic" in the specification of a frailty term in a survreg model.
Thanks for your help - apologies in advance if I'm missing something very
obvious. I'm using R2.13.0 btw, with survival v2.39-9.
Regards
Carina
[[alternative HTML version deleted]]
______________________________________________
R-help@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-help
PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
and provide commented, minimal, self-contained, reproducible code.
