Skip to content

System crash when using surv=T in cph

3 messages · maxb, Uwe Ligges, Frank E Harrell Jr

Original
maxb
# 
Can someone help me. I am very new to R. I am fitting a Cox model using Frank
Harrell's cph as I want to produce a Nomogram. This is what I have done:

Srv<- Surv(time,cens)
f.cox<- cph(Srv~ v1+v2+v3+v4, x=T, y=T, surv=T)

As soon as I press enter, Windows XP crashes. If I remove surv=T, then it
works. I have R version 2.9.0.

Is there a way of displaying the parameter estimates (ie beta coefficients)
and HR when I type 
anova(f.cox) as this only displays the Chi squared and p-values.
Any help or advise drawing a Nomogram will be appreciated.

Thanks in advance

Max
Uwe Ligges
#
maxb wrote:
This is not reproducible for us. Where are the data?
What is cph? I presume the one from package Design?
Have you updates all your packages? Which versions of survival and 
Design are you using?
Is it really Windows that crashes, or just R?


See
http://cran.r-project.org/web/checks/check_results_Design.html
to learn that even the newest version of Design produces WARNINGs.
If you can send reproducible code to let R crash, please send it to the 
package maintainer of the involved packages.

Best,
Uwe Ligges
Frank E Harrell Jr
#
maxb wrote:
Until Design is updated (which will be very soon)
source('http://biostat.mc.vanderbilt.edu/tmp/cphnew.s') after 
library(Design).

anova is not supposed to display parameter estimates.  For one thing, 
there is often more than one parameter associated with a term in the 
model.  Use summary(fit) to get hazard ratios for sensible covariate ranges.

Frank