[R] Measure the frequencies of pairs in a matrix

[Boris Steipe]

Still not sure I understand. But here is what I think you might mean:

# Your data
mat <- structure(c(5, 6, 5, 5, 4, 3, 6, 7, 4, 7, 5, 5, 5, 5, 6, 5, 5,
4, 3, 6, 7, 4, 7, 5, 5, 5, 6, 5, 4, 5, 5, 7, 5, 6, 3, 5, 6, 7,
6, 6, 5, 4, 5, 5, 7, 5, 6, 3, 5, 6, 7, 6), .Dim = c(26L, 2L))

# Create a square matrix with enough space to have an element for each pair. Since
# order is not important, only the upper triangle is used. If the matrix is
# large and sparse, a different approach might be needed.
freq <- matrix(numeric(max(mat) * max(mat)),  nrow = max(mat), ncol = max(mat))

# Loop over your input
for (i in 1:nrow(mat)) {
    # Sort the elements of a row by size.
    x <- sort(mat[i,])
    # Increment the corresponding element of the frequency matrix
    freq[x[1], x[2]] <- freq[x[1], x[2]] + 1
}

freq


Cheers,
B.
 




On Oct 7, 2015, at 1:17 AM, Hermann Norpois <hnorpois at gmail.com> wrote:

> Ok, this was misleading. And was not that important. My result matrix should look like this: 
> 
>   1    2   3   4   5   6   7 ...
> 1 p1 p2
> 2 p
> 3
> 4
> 
> p1 etc are the frequencies of the combinations
> 
> 1 and 2 for instance do not appear in my example. So the values would be zero. Actually, this part is not too important. I would be happy enough to solve the challenge with the frequencies of the pairs.
> Thanks Hermann
> 
> 2015-10-07 2:40 GMT+02:00 Boris Steipe <boris.steipe at utoronto.ca>:
> Since order is not important to you, you can order your pairs (e.g. decreasing) before compiling the frequencies.
> But I don't understand the second part about values "that do not appear in the matrix". Do you mean you want to assess all combinations? If that's the case I would think about a hash table or other indexed data structure, rather than iterating through a matrix.
> 
> 
> B.
> 
> 
> 
> On Oct 6, 2015, at 4:59 PM, Hermann Norpois <hnorpois at gmail.com> wrote:
> 
> > Hello,
> >
> > I have a matrix mat (see dput(mat))
> >
> >> mat
> >      [,1] [,2]
> > [1,]    5    6
> > [2,]    6    5
> > [3,]    5    4
> > [4,]    5    5
> > ....
> >
> > I want the frequencies of the pairs in a new matrix, whereas the
> > combination 5 and 6 is the same as 6 and 5 (see the first two rows of mat).
> > In other words: What is the probability of each combination (each row)
> > ignoring the order in the combination. As a result I would like to have a
> > matrix that includes rows and cols 0, 1, 2 ... max (mat) that do not appear
> > in my matrix.
> >
> > dput (mat)
> > structure(c(5, 6, 5, 5, 4, 3, 6, 7, 4, 7, 5, 5, 5, 5, 6, 5, 5,
> > 4, 3, 6, 7, 4, 7, 5, 5, 5, 6, 5, 4, 5, 5, 7, 5, 6, 3, 5, 6, 7,
> > 6, 6, 5, 4, 5, 5, 7, 5, 6, 3, 5, 6, 7, 6), .Dim = c(26L, 2L))
> >
> > Thanks
> > Hermann
> >
> >       [[alternative HTML version deleted]]
> >
> > ______________________________________________
> > R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> > https://stat.ethz.ch/mailman/listinfo/r-help
> > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> > and provide commented, minimal, self-contained, reproducible code.
> 
> ______________________________________________
> R-help at r-project.org mailing list -- To UNSUBSCRIBE and more, see
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.
>