[R] probelem of function inside function

[Simon Blomberg]

Your function works fine for me (R 1.7.1, Windows 2000):

> library(MASS)
> testdata<-data.frame(cbind(x=c(129,109,52,118,217,278,606,198,99,133),y=c(89,118,99,137,34,14,130,30,131,30)))
> f.fun<-function(var,fdata) {
+       fdata<-data.frame(fdata)
+       fit<-lm(y~1,data=fdata)
+ 
+       f.addterm<-function(x,fit,f.data) {
+ fdata<-f.data
+    fff<-paste('~.',x,sep='+')
+    try(addterm(fit,as.formula(fff))$AIC[2])
+       }
+ 
+       f.addterm(var,fit,fdata)
+  }
> f.fun("x", testdata)
       x 
80.81915 
> 

Simon.

Simon Blomberg, PhD
Depression & Anxiety Consumer Research Unit
Centre for Mental Health Research
Australian National University
http://www.anu.edu.au/cmhr/
Simon.Blomberg at anu.edu.au  +61 (2) 6125 3379


> -----Original Message-----
> From: array chip [mailto:arrayprofile at yahoo.com]
> Sent: Thursday, 26 June 2003 6:48 AM
> To: r-help at stat.math.ethz.ch
> Subject: [R] probelem of function inside function
> 
> 
> Hi,
> 
> I encountered a problem when I am trying to write my
> own function which contains another function. To
> simplify a problem, I tried the following simplified
> function, hope someone can idenfity the problem for
> me.
> 
> I have a simple data frame called "testdata" as
> following:
> 
> >
> testdata<-data.frame(cbind(x=c(129,109,52,118,217,278,606,198,
> 99,133),y=c(89,118,99,137,34,14,130,30,131,30)))
> > testdata
>      x   y 
>  1 129  89
>  2 109 118
>  3  52  99
>  4 118 137
>  5 217  34
>  6 278  14
>  7 606 130
>  8 198  30
>  9  99 131
> 10 133  30
> 
> I have write a simple function (nonsense):
> 
> > f.fun<-function(var,fdata) {
>       fdata<-data.frame(fdata)
>       fit<-lm(y~1,data=fdata)
> 
>       f.addterm<-function(x,fit,f.data) {
> 	fdata<-f.data
>    	fff<-paste('~.',x,sep='+')
>    	try(addterm(fit,as.formula(fff))$AIC[2])
>       }
> 
>       f.addterm(var,fit,fdata)
>  }
> 
> This function simply add the "x" variale to the linear
> regression with intercept only using addterm, and
> return the AIC of the new model. I know I can simply
> do this with stepAIC of MASS, but for my own
> situation, I want to try something more complicated.
> 
> so when I execute the following command, I got an
> error message:
> 
> > f.fun(var='x',fdata=testdata)
> Problem in eval(oc, list()): Object "fdata" not found 
> Use traceback() to see the call stack
> [1] "Problem in eval(oc, list()): Object \"fdata\" not
> found \nUse traceback() to see the call stack"
> attr(, "class"):
> [1] "Error"
> 
> I guess the reason is that within the inside function
> "f.addterm", the dataset "fdata" can't be seen, but I
> don't know why because I think I have passed the
> dataset into the inside function.
> 
> However, if I create a dataset called "fdata" outside
> the function "f.fun" at the very beginning, everything
> will be fine. This is obvious because now "fdata" is a
> global dataset now.
> 
> > fdata=testdata
> > f.fun(var='x',fdata=testdata)
> [1] 80.81915
> 
> 
> Thanks
> 
> Zander.
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
>