[Rd] xtabs to dataframe fails (PR#3754)

Uwe Ligges ligges at statistik.uni-dortmund.de
Sat Aug 16 14:17:33 MEST 2003



daniel.frey at switzerland.org wrote:
> 
> Full_Name: Daniel Frey
> Version: 1.7.1
> OS: Windows 2000
> Submission from: (NULL) (80.254.164.242)
> 
> Generating a data frame out of a xtabs result acts unusual. Take the following
> sample to reproduce it:
> 
>     > a.a <- c("a","a","a","b","b")
> 
>     > a.b <- c("c","c","d","e","f")
> 
>     > a.df <- data.frame(list("A"=a.a,"B"=a.b))
> 
>     > a.x <- xtabs(~A+B,a.df)
> 
>     > a.x
>        B
>     A   c d e f
>       a 2 1 0 0
>       b 0 0 1 1
> 
>     > data.frame(a.x)
>       A B Freq
>     1 a c    2
>     2 b c    0
>     3 a d    1
>     4 b d    0
>     5 a e    0
>     6 b e    1
>     7 a f    0
>     8 b f    1
> 
> I would expect something like a.x itself. Instead I have to give the exact
> bounds, otherwise it doesn't convert properly:
> 
>     > data.frame(a.x[1:2,1:4])
>       c d e f
>     a 2 1 0 0
>     b 0 0 1 1
> 
> I consider this a bug, as consistency of the handling breaks here.
> 
> Daniel Frey


a) data.frame() calls as.data.frame()
b) as.data.frame() uses its method as.data.frame.table(), because
  class(a.x)
  [1] "xtabs" "table"
c) ?as.data.frame.table tells us:
     `as.data.frame.table' is a method for the generic function
     `as.data.frame' to convert the array-based representation of a
     contingency table to a data frame containing the classifying
     factors and the corresponding counts (the latter as component
     `Freq'). This is the inverse of `xtabs'.

So, what is the bug?

I guess you expect something like 
  data.frame(unclass(a.x))

Uwe Ligges



More information about the R-devel mailing list