R help - Nov 2003 - converting column to factor *within* a data frame

Hi all,

I repeatedly encounter the following problem: After importing a data set 
into a data frame, I wish to set a column with numeric values to be a 
factor, but can't figure out how to do this. Also, I do not wish to 
write as.factor(x) all the time. I can create a new vector with x <- 
factor(x), but the new vector resides outside the attached data frame.

Pascal

 > attach(ngrad)
 > is.factor(STNW)
[1] FALSE

 > ngrad$STNW<-factor(STNW)  ## doesn't work
 > is.factor(STNW)
[1] FALSE

 > is.factor(STNW) <- T    ## doesn't work either
Error: couldn't find function "is.factor<-"

On Wed, 5 Nov 2003, Pascal A. Niklaus wrote:
> Hi all,
> 
> I repeatedly encounter the following problem: After importing a data set 
> into a data frame, I wish to set a column with numeric values to be a 
> factor, but can't figure out how to do this. Also, I do not wish to 
> write as.factor(x) all the time. I can create a new vector with x <- 
> factor(x), but the new vector resides outside the attached data frame.
> 
> Pascal
> 
>  > attach(ngrad)
>  > is.factor(STNW)
> [1] FALSE
> 
>  > ngrad$STNW<-factor(STNW)  ## doesn't work
>  > is.factor(STNW)
> [1] FALSE
It does work.  It changes ngrad, and not the copy you attached.

ngrad$STNW<-factor(ngrad$STNW)
attach(ngrad)

is the correct sequence.

-- 
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

Your problem is with scoping, not the conversion per se:
>  > attach(ngrad)
>  > is.factor(STNW)
> [1] FALSE
At this moment, STNW is the same as ngrad$STNW
>  > ngrad$STNW<-factor(STNW)  ## doesn't work
Yes it does work, try looking at is.factor(ngrad$STNW)
>  > is.factor(STNW)
> [1] FALSE
After you assign to ngrad$STNW, it is no longer the same
thing as the attached STNW. You would need to detach
and re-attach ngrad for this to be so. There's no
automatic synchronisation between the attached STNW 
and ngrad$STNW; changing one will not change the other.

My advice is: never use attach() if you can help it. It's an 
accident waiting to happen. Get used to typing
dataFrame$varname instead of just varname - that way
you will always get what you expect. Or use with() instead 
of attach() in almost every case. 

HTH  
 
Simon Fear 
Senior Statistician 
Syne qua non Ltd 
Tel: +44 (0) 1379 644449 
Fax: +44 (0) 1379 644445 
email: Simon.Fear at synequanon.com 
web: http://www.synequanon.com 
  
Number of attachments included with this message: 0 
  
This message (and any associated files) is confidential and\...{{dropped}}

Pascal A. Niklaus <Pascal.Niklaus at unibas.ch>
wrote:
> I repeatedly encounter the following problem: After importing a data set 
> into a data frame, I wish to set a column with numeric values to be a 
> factor, but can't figure out how to do this. Also, I do not wish to 
> write as.factor(x) all the time. I can create a new vector with x <- 
> factor(x), but the new vector resides outside the attached data frame.
> 
> Pascal
> 
> > attach(ngrad)
> > is.factor(STNW)
> [1] FALSE
> 
> > ngrad$STNW<-factor(STNW)  ## doesn't work
> > is.factor(STNW)
> [1] FALSE
That command checked the attached dataset, which has not been
modified by the previous command. You seem to be confused by what
attach() does. Read ?attach.
> data <- data.frame(a=1:3, b=4:6)
> data
  a b
1 1 4
2 2 5
3 3 6
> attach(data)
> is.factor(a)
[1] FALSE
> data$a <- as.factor(a)
> is.factor(a)
[1] FALSE
> is.factor(data$a)
[1] TRUE
> detach(data)
> attach(data)
> is.factor(a)
[1] TRUE

-- 
Philippe