R help - Aug 2010 - rpart - interpretation of results of tree on survival data

Hi All,

I am fitting a tree to censored survival data using the rpart package and
wanted to better understand the results.

I am trying to interpret the output from the tree. I am interested in
understanding what "yval" is for a survival tree. I see in the output
of
summary, the phrase "estimated rate". The estimated rate is 1 for the
entire
tree, and more of less for each node. So, I am guessing it is some function
of the event rate in each node compared to the full tree. But then I saw a
node with no events a  non-null estimated rate, that is a little confusing.

I have the document called "An Introduction to Recursive Partitioning Using
the RPART Routines", while it does mention "exponential scaling",
the
explanation rather brief. I will be very grateful if someone could guide me
to some more documentation.

Here is some reproducible code:
library(rpart)
bladder1 <- bladder[bladder$enum < 5, ]
fit <- rpart(Surv(stop, event) ~ rx + size + number, bladder1)

print(fit)
summary(fit)
plot(fit, uniform=T, branch=.4, compress=T)
text(fit, use.n=T)

Here is an excerpt from summary(fit):
Node number 56: 8 observations
  events=0,  estimated rate=0.1858292 , mean deviance=0.2035427

Thanks in advance,
Ritwik
> R.Version()
$platform
[1] "i386-pc-mingw32"

$arch
[1] "i386"

$os
[1] "mingw32"

$system
[1] "i386, mingw32"

$status
[1] ""

$major
[1] "2"

$minor
[1] "10.0"

$year
[1] "2009"

$month
[1] "10"

$day
[1] "26"

$`svn rev`
[1] "50208"

$language
[1] "R"

$version.string
[1] "R version 2.10.0 (2009-10-26)"


-- 
Ritwik Sinha
ritwik.sinha@gmail.com | +12038160799 | http://ritwik.sinha.googlepages.com/

	[[alternative HTML version deleted]]