[R] command in survival package

Marc Schwartz MSchwartz at mn.rr.com
Sat Jan 21 01:49:02 CET 2006


On Fri, 2006-01-20 at 15:27 -0800, Thomas Lumley wrote:
> On Fri, 20 Jan 2006, Linda Lei wrote:
> 
> > Hi there,
> >
> >
> >
> > I have a question about one command sentence when I follow the example
> > in the book of "Survival analysis in S":
> >
> > > aml1<-aml[aml$group==1]
> 
> If this is really what the book says you should probably complain to the 
> author.  However, if it says
>    aml1<-aml[aml$group==1,]
> then you show type that instead.
> 
> It is hard to be sure, because you don't say where the `aml' data set 
> comes from.  It can't be the one in the survival package as this doesn't 
> have a variable called "group"
> 
> >
> > Thus, I couldn't keep going on the next command:
> >
> > esf.fit<-survfit(Surv(aml1,status)~1).
> >
> 
> This looks implausible as well. I would have expected something like
>    esf.fit<-survfit(Surv(time,status)~1, data=aml1)


Correct on both accounts Thomas. The text in the book on page 26 appears
as Linda posted.

However, there is an errata document here:

http://www.crcpress.com/e_products/downloads/download.asp?cat_no=C4088

which shows the corrected text as:

aml1 <- aml[aml$group==1,] # Maintained group only

esf.fit <- survfit(Surv(aml1$weeks,status) ~ 1)


HTH,

Marc Schwartz




More information about the R-help mailing list