[R] Most often pairs of chars across grouping variable

[Marc Schwartz]

on 07/29/2008 09:51 AM svga at arcor.de wrote:
> Hi list,
> 
> is there a package or function to compute the frequencies of pairs of
> chars in a variable across a grouping variable? Eg:
> 
> 
> d <- data.frame(ID=gl(2,3), F=c("A","B","C","A","C","D"))
>> d
> ID F 1  1 A 2  1 B 3  1 C 4  2 A 5  2 C 6  2 D
> 
> 
> Now I want to summarize the frequencies of all pairs A-B, A-C, A-D,
> B-C, B-D, C-D across ID:
> 
> A B C D A  - 1 2 1 B  - - 1 0 C  - - - 1
> 
> 
> here, the combination A-C is most frequent. The real problem behind
> that is that 'F' codes diagnoses and I search for the most often
> pairs of diagnoses.
> 
> Thanks, Sven

I suspect that there might be something over in Bioconductor, but here 
is one approach:

 > table(data.frame(t(do.call(cbind,
                      tapply(d$F, d$ID,
                             function(x) combn(as.character(x), 2))))))
    X2
X1  B C D
   A 1 2 1
   B 0 1 0
   C 0 0 1


See ?combn to create the initial pairs from the data. This is done on a 
per ID basis using tapply. The result is transposed into a data frame 
and then table() is used to create the cross tabulation of the results.

HTH,

Marc Schwartz