[R] RPART - printing full splitting rule number on tree plot

[Prof Brian Ripley]

On Mon, 14 Dec 2009, Rebecca O'Leary wrote:

> Dear R-users
>
>
>
> I am using RPART package to get regression trees.  However having 
> trouble getting the text function to put the full splitting rule 
> number on the plot, instead to puts it in scientific notation. 
> When a covariate has 1e4 or greater number of digits then the 
> splitting rule number displayed on the plot is in scientific 
> notation.  But print.rpart displays the splitting rules in full.  I 
> have tried using digits option in text function but this only alters 
> digits of the mean displayed at terminal nodes.  How do I get the 
> full splitting rule number displayed on the tree plot?

You change rpart:::labels.rpart, which uses a 'g' format and hence 
uses scientific notation if it is more compact.

>
> Below is an example.
>
>
>
> library(rpart)
>
> temp <- as.data.frame(cbind(kyphosis, Start2=kyphosis$Start))
>
>
>
> temp$Start2 <- temp$Start2+10000
>
>
>
> fit1 <- rpart(Age~ Kyphosis +Number + Start2, data=temp)
>
> plot(fit1)
>
> text(fit1, use.n=TRUE, digits=5)
>
> Thanks
>
>
> Dr Rebecca O'Leary, PhD
>
> Biostatistician
> Senior Research Officer
> UWA Centre for Child Health Research
> Telethon Institute for Child Health Research, 100 Roberts Rd, Subiaco
> Perth WA 6008 Australia
> Email: roleary at ichr.uwa.edu.au
> Tel: +61-8-9489 7745
> Fax: +61-8-9489 7700
>
> 	[[alternative HTML version deleted]]

Please send properly formatted text mail without all the blank lines.

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595