0

I want to extend the solution of this post where @AnandaMahto gave a very elegant solution to my problem.

For this new function, I'd like that if there are several times the same species in the same house, it will count only one observation. One house with two cats and one rat does not create two observations between cat and rat but only one (As shown below)

In this example, there are two rats in the house number 4. As already said, I do not want to consider two observations between rat and cat and between spider and rat but only one observation between rat and cat and one observation between spider and rat

houses = c(1,1,2,3,4,4,4,4,5,6,5)
animals = c('cat','dog','cat','dog','rat', 'cat', 'spider', 'rat', 'cat', 'cat', 'rat')

@AnandaMahto's solution would return this:

        dog rat spider
   cat   1   3      1
   dog       0      0
   rat              2

But I would like to get this:

        dog rat spider
   cat   1   2      1
   dog       0      0
   rat              1
Community
  • 1
  • 1
Remi.b
  • 17,389
  • 28
  • 87
  • 168

1 Answers1

2

Make all values > 0 from table equal to "1" before using crossprod:

(table(houses, animals) > 0) *1
#       animals
# houses cat dog rat spider
#      1   1   1   0      0
#      2   1   0   0      0
#      3   0   1   0      0
#      4   1   0   1      1
#      5   1   0   1      0
#      6   1   0   0      0

out <- crossprod((table(houses, animals) > 0) *1)
out[lower.tri(out, diag=TRUE)] <- NA
as.table(out)
#         animals
# animals  cat dog rat spider
#   cat          1   2      1
#   dog              0      0
#   rat                     1
#   spider

To get to the desired output, since we know that the first column and the last row will be empty, and since you already figured out on your own that as.table would take care of not printing the NA values, continuing from above, you can do:

out <- as.table(out[-nrow(out), -1])
out
#        animals
# animals dog rat spider
#     cat   1   2      1
#     dog       0      0
#     rat              1
A5C1D2H2I1M1N2O1R2T1
  • 190,393
  • 28
  • 405
  • 485