[Rd] rpart package, text function, and round of class counts

[Achim Zeileis]

On Sun, 4 Mar 2012, yindalon wrote:

> I run the following code:
>
> library(rpart)
> data(kyphosis)
> fit <- rpart(Kyphosis ~ ., data=kyphosis)
> plot(fit)
> text(fit, use.n=TRUE)
>
> The text labels represent the count of each class at the leaf node.
> Unfortunately, the numbers are rounded and in scientific notation rather
> than the exact number of examples sorted by that node in each class.

You probably have a getOption("digits") of 4 or lower. text.rpart uses 
getOption("digits") - 3 as the default which then means only 1 significant 
digit and hence it rounds and uses scientific notation. Using

text(fit, use.n = TRUE, digits = 3)

should do the trick. Maybe adding setting xpd = TRUE in addition helps in 
avoiding clipping of some labels.

Also, I would recommend to use

library("partykit")
plot(as.party(fit))

for visualization which uses a display like for the ctree() function (also 
mentioned on the web page you quote below).

> The plot is supposed to look like
> http://www.statmethods.net/advstats/images/ctree.png as per
> http://www.statmethods.net/advstats/cart.html.
>
> I'm running 2.14.1 on a mac.
>
> Can anyone verify or point out if I am doing something obviously wrong for
> displaying the counts rounded and in scientific notation rather than the
> true counts in each class at each node?
> Thanks.
>
> --
> View this message in context: http://r.789695.n4.nabble.com/rpart-package-text-function-and-round-of-class-counts-tp4444576p4444576.html
> Sent from the R devel mailing list archive at Nabble.com.
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>