I've figured it out by repeatedly testing. It is to use a type='term'
statement, just as used in gam.
sorry to bother.
On 2/19/08, gallon li <[EMAIL PROTECTED]> wrote:
>
> Thanks a lot, Prof Lumley.
>
> Now I can fit a model like
>
> coxfit=coxpy((time,censor)~pspline(x1)+x2+x3)
>
> but I am not sure how to extract the estimated function for x1 alone. I
> tried to use predict function but couldn't find appropriate option to do
> this.
>
> if i only have one covariates, then the example in help manule can be used
> as
> plot(x1, predict(coxfit))
>
> but with more than 1 predictor, i am not sure how to select the one i
> want.
>
>
> On 2/19/08, Thomas Lumley <[EMAIL PROTECTED]> wrote:
> >
> > On Mon, 18 Feb 2008, gallon li wrote:
> >
> > > i am trying to fit a survival regression model (cox model or
> > parametric
> > > model) in R by including the covariate effects as a function m(x)
> > instead of
> > > just beta*x. is it possible to fit such a model? can someone recommend
> > some
> > > reference? I searched but only found a package called addreg where
> > > the hazard is actually modeled additively. That is not what i want.
> >
> > The survival package has pspline() for this purpose (assuming that by
> > 'non-parametric' you mean a flexible smooth curve)
> >
> > -thomas
> >
> > Thomas Lumley Assoc. Professor, Biostatistics
> > [EMAIL PROTECTED] University of Washington, Seattle
> >
>
>
[[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.
