[R] Problem with rpart

[Andrew Zachary]

Here is an example (though the data are too large to send ). The dataset
is (6530 x 15). Predictors are continuous N(0,1). Trying to build a
regression tree.

 fit <- rpart( y ~ x1 + x2 + x3 + x4 + x5 + x6 + x7 + x8 + x9 + x10 +
x11 + x12 + x13 + x14, data=my.data.set, weights=wts )

And the output:

 summary( fit )
Call:
rpart(formula = y ~ x1 + x2 + x3 + x4 + x5 + x6 + x7 + x8 + x9 + x10 +
x11 + x12 + x13 + x14, data =my.data.set, weights = wts)
  n= 6530 

  CP nsplit rel error
1 NA      0        NA

Node number NA: NA observationsError in if (ff$complexity[i] < cp ||
is.leaf[i]) cat("\n") else cat(",    complexity param=",  : 
        missing value where TRUE/FALSE needed


If I run this using a subset of 900 points, everything is fine.
Similarly, if I run it using 1100 points, it dies. There are no missing
values in the dataset. Is this simply a case where I should decrease cp?

Regards,
Andrew

-----Original Message-----
From: Torsten Hothorn [mailto:Torsten.Hothorn at rzmail.uni-erlangen.de] 
Sent: Tuesday, September 19, 2006 4:45 PM
To: Andrew Zachary
Cc: r-help at stat.math.ethz.ch
Subject: Re: [R] Problem with rpart


On Tue, 19 Sep 2006, Andrew Zachary wrote:

> Not sure if anyone has posted on this problem ... I want to use rpart 
> to build a binary tree on a relatively large dataset with ~1400 data 
> points and 15 predictors. But I've noticed that rpart fails almost 
> immediately in the call to C_s_to_rp, as that code returns nonsense. 
> Looking at the code itself isn't terribly helpful, and there don't 
> seem to be any hard limits coded anywhere. Does anyone have a 
> suggestion for what might be going on?
>

Andrew,

you need to give an _executable_ example illustrating your problem. What
means `nonsense'?

Best,

Torsten

> Thanks in advance for you help
> Andrew Zachary
>
> ----
> Wetherby Partners LLC believes the information provided herein is
reliable. While every care has been taken to ensure accuracy, the
information is furnished to the recipients with no warranty as to the
completeness and accuracy of its contents and on condition that any
errors or omissions shall not be made the basis for any claim, demand or
cause for action.
> The information in this email is intended only for the\ > ...{{dropped}}