R help - May 2007 - Error in if (!length(fname) || !any(fname == zname)) { :

Dear R users,

I tried to fit a cox proportional hazard model to get estimation of stratified
survival probability. my R code is as follows:

cph(Surv(time.sur, status.sur)~ strat(colon[,13])+colon[,18]
+colon[,20]+colon[,9], surv=TRUE)
Error in if (!length(fname) || !any(fname == zname)) { : 
        missing value where TRUE/FALSE needed
Here colon[,13] is the one that I want to stratify and the others are all
coefficients. How can I solve this problem?  Thanks a lot!!!!!

Grace

       
---------------------------------


	[[alternative HTML version deleted]]

For me, the simplest way to find, what is wrong, would be tracing the R code:

library(debug)
mtrace(cph)
cph(Surv(time.sur, status.sur)~ strat(colon[,13])+colon[,18]
+colon[,20]+colon[,9], surv=TRUE)

... then find the place of the error and analyze how to adjust the function
call arguments to avoid the error and get the desired result.


hongyuan cao wrote:
> 
> I tried to fit a cox proportional hazard model to get estimation of
> stratified survival probability. my R code is as follows:
> 
> cph(Surv(time.sur, status.sur)~ strat(colon[,13])+colon[,18]
> +colon[,20]+colon[,9], surv=TRUE)
> Error in if (!length(fname) || !any(fname == zname)) { : 
>         missing value where TRUE/FALSE needed
> Here colon[,13] is the one that I want to stratify and the others are all
> coefficients. How can I solve this problem?  Thanks a lot!!!!!
> 
> Grace
> 
-- 
View this message in context:
http://www.nabble.com/Error-in-if-%28%21length%28fname%29-%7C%7C-%21any%28fname-%3D%3D-zname%29%29-%7B-%3A-tf3691016.html#a10319988
Sent from the R help mailing list archive at Nabble.com.

Vladimir Eremeev wrote:
> 
> For me, the simplest way to find, what is wrong, would be tracing the R
> code:
> 
> library(debug)
> mtrace(cph)
> cph(Surv(time.sur, status.sur)~ strat(colon[,13])+colon[,18]
> +colon[,20]+colon[,9], surv=TRUE)
> 
> ... then find the place of the error and analyze how to adjust the
> function call arguments to avoid the error and get the desired result.
> 
and then, switch off the  tracing
mtrace(cph,FALSE)
or
mtrace.off()
-- 
View this message in context:
http://www.nabble.com/Error-in-if-%28%21length%28fname%29-%7C%7C-%21any%28fname-%3D%3D-zname%29%29-%7B-%3A-tf3691016.html#a10320118
Sent from the R help mailing list archive at Nabble.com.

hongyuan cao wrote:
> Dear R users,
> 
> I tried to fit a cox proportional hazard model to get estimation of
stratified survival probability. my R code is as follows:
> 
> cph(Surv(time.sur, status.sur)~ strat(colon[,13])+colon[,18]
+colon[,20]+colon[,9], surv=TRUE)
> Error in if (!length(fname) || !any(fname == zname)) { : 
>         missing value where TRUE/FALSE needed
> Here colon[,13] is the one that I want to stratify and the others are all
coefficients. How can I solve this problem?  Thanks a lot!!!!!
> 
> Grace
The Design package does not like you to have complex variable names like 
that, and in general storing your data in a matrix when you want to 
treat columns as separate variables is not the best approach.  I would 
use something like

S <- with(d, Surv(  ) )   # d = data frame
h <- as.data.frame(colon)  # assumes colon is a matrix;make sure it had 
column names
cph(S ~ strat(v1)+v2+v3+v4, data=h)

-- 
Frank E Harrell Jr   Professor and Chair           School of Medicine
                      Department of Biostatistics   Vanderbilt University