[R] creating objects of class "xtabs" "table" in R

[Marc Schwartz]

On Fri, 2007-10-05 at 13:09 -0700, Deepayan Sarkar wrote:
> On 10/5/07, born.to.b.wyld at gmail.com <born.to.b.wyld at gmail.com> wrote:
> > I have an application that would generate a cross-tabulation in array
> > format in R. In particular, my application would give me a result
> > similar to that of :
> >
> > array(5,c(2,2,2,2,2))
> >
> > The above could be seen as a cross-tabulation of 5 variables with 2
> > levels each (could be 0 and 1). In this case, the data were such that
> > each cell has exactly 5 observations. I
> >
> > Now, I want the output to look like the output of 'xtabs' utility, so
> > that I can use this output inside 'loglm(MASS)'. In particular, I want
> > to add (variable) names to each dimension and indicate the levels that
> > correspond to each cell. The output from 'xtabs' for a data set of
> > this kind would look like:
> 
> Simplifying your example:
> 
> > foo <- array(5,c(2,2), dimnames = list(xx = c("x=0", "x=1"), yy = c("A", "B")))
> > foo
>      yy
> xx    A B
>   x=0 5 5
>   x=1 5 5
> 
> You can also do this in two steps:
> 
> foo <- array(5,c(2,2))
> dimnames(foo) <-  list(xx = c("x=0", "x=1"), yy = c("A", "B"))
> 
> [...]
> 
> > Now, I do not generate my output using the 'xtabs' utility. In fact,
> > my simulations would generate the cross-table directly and not the
> > dataset.
> >
> > Can anyone help? R examples have some reference to the 'dimnames'
> > attribute, but I am not exactly sure.
> > Also, is there an R function that could do the exact opposite of
> > 'xtabs', that is, may be generate a data frame given its cross-table?
> 
> Sort of (there is only one column for each combination, giving `frequencies'):
> 
> > as.data.frame.table(foo)
>    xx yy Freq
> 1 x=0  A    5
> 2 x=1  A    5
> 3 x=0  B    5
> 4 x=1  B    5
> 
> If foo has class "table", then as.data.frame(foo) would also work.

In follow up to Deepayan's reply, I have posted this previously, but
here is a function that will take the result of applying
as.data.frame.table() to a table and generate the raw data. For example,
using Deepayan's table above:


> foo
     yy
xx    A B
  x=0 5 5
  x=1 5 5

f.dft <- as.data.frame.table(foo)

> f.dft
   xx yy Freq
1 x=0  A    5
2 x=1  A    5
3 x=0  B    5
4 x=1  B    5


Here is the function, called expand.dft():


expand.dft <- function(x, na.strings = "NA", as.is = FALSE, dec = ".")
{
  DF <- sapply(1:nrow(x), function(i) x[rep(i, each = x$Freq[i]), ],
               simplify = FALSE)

  DF <- subset(do.call("rbind", DF), select = -Freq)

  for (i in 1:ncol(DF))
  {
    DF[[i]] <- type.convert(as.character(DF[[i]]),
                            na.strings = na.strings,
                            as.is = as.is, dec = dec)
                                           
  }
    
  DF
}          


Now, applying that to 'f.dft' from above:

> expand.dft(f.dft)
     xx yy
1   x=0  A
1.1 x=0  A
1.2 x=0  A
1.3 x=0  A
1.4 x=0  A
2   x=1  A
2.1 x=1  A
2.2 x=1  A
2.3 x=1  A
2.4 x=1  A
3   x=0  B
3.1 x=0  B
3.2 x=0  B
3.3 x=0  B
3.4 x=0  B
4   x=1  B
4.1 x=1  B
4.2 x=1  B
4.3 x=1  B
4.4 x=1  B


and of course:

> table(expand.dft(f.dft))
     yy
xx    A B
  x=0 5 5
  x=1 5 5


HTH,

Marc Schwartz