R help - Apr 2009 - argument 'exclude' in function xtabs

Dear all

I was willing to use argument 'exclude' in function xtabs to remove some
levels of factors (xtabs help page says '"exclude: a vector of values
to be
excluded when forming the set of levels of the classifying factors").


I tried:
> mydata <- data.frame(
     treatment = c("B", "A", "C", "C",
"B", "B", "C", "A", "B",
"B", NA,
"C"),
     surv = c("YES", "NO", "YES", "YES",
"NO", "NO", "NO", "YES",
"YES",
"NO", "NO", NA)
    )
> levels(mydata$treatment)
[1] "A" "B" "C"
> # try to remove level "B" in variable treatment
> xtabs(formula = ~ treatment + surv, data = mydata, exclude = "B")
        surv
treatment NO YES
       A  1   1
       B  3   2
       C  1   2



One alternative is to do:
> xtabs(formula = ~ treatment + surv, data = mydata[mydata$treatment !=
"B",
], drop.unused.levels = TRUE)
        surv
treatment NO YES
       A  1   1
       C  1   2



But I don't understand why I cannot remove "B" directly with
argument
exclude.

Any help is welcome

Regards

Matthieu





My version on Windows XP

platform       i386-pc-mingw32
arch           i386
os             mingw32
system         i386, mingw32
status
major          2
minor          9.0
year           2009
month          04
day            17
svn rev        48333
language       R
version.string R version 2.9.0 (2009-04-17)





------------------
Matthieu Lesnoff
CIRAD
Bamako, Mali
E-mail: matthieu.lesnoff@cirad.fr

	[[alternative HTML version deleted]]

> I was willing to use argument 'exclude' in function xtabs to remove
some
> levels of factors (xtabs help page says '"exclude: a vector of
values to
> be excluded when forming the set of levels of the classifying
factors").

I think I see what's happening, and it's a little confusing to me, too.
If your classifying factor to xtabs is not
actually a factor, but say, a character vector, it will be converted to a
factor
while using the "exclude" argument.  However, if it already is a
factor,
this does not happen.
So, for example, contrast:

trt.char <-
c("A","B","C","A","B","C")
xtabs(~trt.char, exclude = "B")

with

trt.fac <-
factor(c("A","B","C","A","B","C"))
xtabs(~trt.fac, exclude = "B")

Hope that helps,
Erik Iverson

Ok thanks Erik

Indeed, "exclude" in "xtabs" seems working whith character
vectors:
> x <- c(rep(c("A","B","C"), 2))
> x
[1] "A" "B" "C" "A" "B"
"C"
> xtabs(~ x, exclude = "B")
x
A C
2 2

and not directly with factors:
> x <- factor(rep(c("A","B","C"), 2))
> x
[1] A B C A B C
Levels: A B C
> levels(x)
[1] "A" "B" "C"
> xtabs(~ x, exclude = "B")
x
A B C
2 2 2

while function "table" does it:
> x <- factor(rep(c("A","B","C"), 2))
> x
[1] A B C A B C
Levels: A B C
> table(x, exclude = "B")
x
A C
2 2

However, one point remains confusing for me in function "table". When
I add
a NA in the vector, it still works:
> x <- factor(c(rep(c("A","B","C"), 2), NA))
> x
[1] A    B    C    A    B    C    <NA>
Levels: A B C
> table(x, exclude = "B")
x
A C
2 2

but not anymore when I use argument "exclude = NULL" in function
"factor":
> x <- factor(c(rep(c("A","B","C"), 2), NA),
exclude = NULL)
> x
[1] A    B    C    A    B    C    <NA>
Levels: A B C <NA>
> table(x, exclude = "B")
x
  A    B    C <NA>
  2    2    2    1

Finally, if I remove NA, it works again:
> x <- factor(rep(c("A","B","C"), 2),
exclude = NULL)
> x
[1] A B C A B C
Levels: A B C
> table(x, exclude = "B")
x
A C
2 2

it looks like "exclude" does not work when NA is a level




Regards

Matthieu
















> -----Message d'origine-----
> De : Erik Iverson [mailto:iverson@biostat.wisc.edu]
> Envoyé : 24 April 2009 19:13
> À : Matthieu Lesnoff
> Cc : r-help@r-project.org
> Objet : Re: [R] argument 'exclude' in function xtabs
>
>
> > I was willing to use argument 'exclude' in function xtabs to
remove
> > some levels of factors (xtabs help page says '"exclude: a
vector of
> > values to be excluded when forming the set of levels of the
> classifying factors").
>
>
> I think I see what's happening, and it's a little confusing
> to me, too.
> If your classifying factor to xtabs is not actually a factor,
> but say, a character vector, it will be converted to a factor
> while using the "exclude" argument.  However, if it already
> is a factor, this does not happen.
> So, for example, contrast:
>
> trt.char <-
c("A","B","C","A","B","C")
> xtabs(~trt.char, exclude = "B")
>
> with
>
> trt.fac <-
factor(c("A","B","C","A","B","C"))
> xtabs(~trt.fac, exclude = "B")
>
> Hope that helps,
> Erik Iverson
>
	[[alternative HTML version deleted]]