R help - Aug 2003 - Error in calling stepAIC() from within a function

Hi,
 
 I am experiencing a baffling behaviour of stepAIC(),
 and I hope to get any advice/help on what went wrong
or I'd missed. I greatly appreciate any advice given.
 
 I am using stepAIC() to, say, select a model via
 stepwise selection method.
 
 R Version : 1.7.1
 Windows ME
 
 Many thanks and best regards,
Siew-Leng
 
 
 ***Issue :
 
 When stepAIC() is placed within a function, it seems
 that stepAIC() cannot detect the data matrix, and
 the
 program is halted as a result. However, when the
 same
 codes are copied, paste and run in R workspace,
 stepAIC can execute and R is able to produce the
 desired output.
 
 ***Actions taken :
 I had tried to look into and manipulate environments
 and formals(f), but with not much luck.
 
 
 ***Code snippets (an example):
 
   library(MASS)
   library(nls)
 
   Data<-data.frame(matrix(rnorm(120),ncol=6))
   colnames(Data)<-c("Y",paste("X",1:5,sep=""))
 
   f<-function(A)
   {
   a <-glm(Y~X1+X2+X3, data=A)
   b

<-stepAIC(a,scope=list(upper=~X1+X2+X3+X4+X5,lower=~1),direction="both",trace=FALSE)
   b
   }
 

  f(Data) 
 
 
 R gives the following error :
 
 > f(Data)
 Error in model.frame.default(formula = Y ~ X1 + X3,
 data = A, drop.unused.levels = TRUE) : 
         Object "A" not found
 
 
 However, when the same codes are copied, paste and
 run
 in the workspace, stepAIC() can run, and R is able
 to
 produce the desired output :
 
 A <- Data
 a <-glm(Y~X1+X2+X3, data=A)
 b

<-stepAIC(a,scope=list(upper=~X1+X2+X3+X4+X5,lower=~1),direction="both",trace=FALSE)
 b
 
> b
 
> Call:  glm(formula = Y ~ X4, data = A) 
> 
> Coefficients:
> (Intercept)           X4  
>      0.1942      -0.3314  
> 
> Degrees of Freedom: 19 Total (i.e. Null);  18
> Residual
> Null Deviance:      20.51 
> Residual Deviance: 17.95        AIC: 60.6 
> > 

__________________________________

Yahoo! SiteBuilder - Free, easy-to-use web site design software

Your example works for me in R-devel, and in R 1.7.1 using step()
rather than stepAIC().

On Sun, 3 Aug 2003, Siew Leng TENG wrote:
> Hi,
>  
>  I am experiencing a baffling behaviour of stepAIC(),
>  and I hope to get any advice/help on what went wrong
> or I'd missed. I greatly appreciate any advice given.
>  
>  I am using stepAIC() to, say, select a model via
>  stepwise selection method.
>  
>  R Version : 1.7.1
>  Windows ME
>  
>  Many thanks and best regards,
> Siew-Leng
>  
>  
>  ***Issue :
>  
>  When stepAIC() is placed within a function, it seems
>  that stepAIC() cannot detect the data matrix, and
>  the
>  program is halted as a result. However, when the
>  same
>  codes are copied, paste and run in R workspace,
>  stepAIC can execute and R is able to produce the
>  desired output.
>  
>  ***Actions taken :
>  I had tried to look into and manipulate environments
>  and formals(f), but with not much luck.
>  
>  
>  ***Code snippets (an example):
>  
>    library(MASS)
>    library(nls)
>  
>    Data<-data.frame(matrix(rnorm(120),ncol=6))
>   
colnames(Data)<-c("Y",paste("X",1:5,sep=""))
>  
>    f<-function(A)
>    {
>    a <-glm(Y~X1+X2+X3, data=A)
>    b
> 
>
<-stepAIC(a,scope=list(upper=~X1+X2+X3+X4+X5,lower=~1),direction="both",trace=FALSE)
>    b
>    }
>  
> 
>   f(Data) 
>  
>  
>  R gives the following error :
>  
>  > f(Data)
>  Error in model.frame.default(formula = Y ~ X1 + X3,
>  data = A, drop.unused.levels = TRUE) : 
>          Object "A" not found
>  
>  
>  However, when the same codes are copied, paste and
>  run
>  in the workspace, stepAIC() can run, and R is able
>  to
>  produce the desired output :
>  
>  A <- Data
>  a <-glm(Y~X1+X2+X3, data=A)
>  b
> 
>
<-stepAIC(a,scope=list(upper=~X1+X2+X3+X4+X5,lower=~1),direction="both",trace=FALSE)
>  b
>  
> > b
>  
> > Call:  glm(formula = Y ~ X4, data = A) 
> > 
> > Coefficients:
> > (Intercept)           X4  
> >      0.1942      -0.3314  
> > 
> > Degrees of Freedom: 19 Total (i.e. Null);  18
> > Residual
> > Null Deviance:      20.51 
> > Residual Deviance: 17.95        AIC: 60.6 
> > > 
> 
> 
> __________________________________
> 
> Yahoo! SiteBuilder - Free, easy-to-use web site design software
> 
> ______________________________________________
> R-help at stat.math.ethz.ch mailing list
> https://www.stat.math.ethz.ch/mailman/listinfo/r-help
> 
-- 
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