[R] using cut on matrices

[Spencer Graves]

Peter's solution led me to an apparent bug in R.

 > a <- 1:9
 > cut(a)
Error in cut.default(a) : Argument "breaks" is missing, with no default

######################
That didn't work, so I read the documentation, found that a second 
argument was required.  Result:
 > cut(a, 2)
[1] (0.992,5] (0.992,5] (0.992,5] (0.992,5] (5,9.01]  (5,9.01]  (5,9.01]
[8] (5,9.01]  (5,9.01]
Levels: (0.992,5] (5,9.01]

##### LOOK VERY CAREFULLY:
##### R 1.6.2 coded 5 as (5, 9.01].
##### (I know I need to upgrade to R 1.7.1.)
##### S-Plus 6.1 for Windows 2000 produced the following:
 > cut(a, 2)
[1] 1 1 1 1 1 2 2 2 2
attr(, "levels"):
[1] "0.92+ thru 5.00" "5.00+ thru 9.08"

##### NOTE:  S-Plus 6.1 did it correctly.
##### I'm sorry to report an error without a fix,
##### but I'm out of time for this now.

##### Before I found the bug, I wrote a function
##### to retain dimnames:

Cut <-
  function(a, ...){
   ac <- cut(a, ...)
   if(is.array(a)){
    dim(ac) <- dim(a)
    dimnames(ac) <- dimnames(a)
   }
   else names(ac) <- names(a)
   ac
}

 > Cut(a, 2)
[1] (0.992,5] (0.992,5] (0.992,5] (0.992,5] (5,9.01]  (5,9.01]  (5,9.01]
[8] (5,9.01]  (5,9.01]
Levels: (0.992,5] (5,9.01]

##### That worked fine.  What about a vector with names?
 > a1 <- a
 > names(a1) <- letters[1:9]
 > Cut(a1, 2)
[1] (0.992,5] (0.992,5] (0.992,5] (0.992,5] (5,9.01]  (5,9.01]  (5,9.01]
[8] (5,9.01]  (5,9.01]
Levels: (0.992,5] (5,9.01]

##### What happened to the names?
 > names(Cut(a1,2))
[1] "a" "b" "c" "d" "e" "f" "g" "h" "i"

##### The names were there, but R chose not to display them.
##### S-Plus 6.1 under Win2000 produced the following:
 > Cut(a1, 2)
  a b c d e f g h i
  1 1 1 1 1 2 2 2 2
attr(, "levels"):
[1] "0.92+ thru 5.00" "5.00+ thru 9.08"

##### The names appear with codes and a translate table
##### Something similar happens with an array:
##### In R 1.6.2:
 > a2 <- a
 > dim(a2) <- c(3,3)
 > dimnames(a2) <- list(LETTERS[1:3], c("ab","bc","cd"))
 > Cut(a2, 2)
[1] (0.992,5] (0.992,5] (0.992,5] (0.992,5] (5,9.01]  (5,9.01]  (5,9.01]
[8] (5,9.01]  (5,9.01]
Levels: (0.992,5] (5,9.01]
 > dimnames(Cut(a2,2))
[[1]]
[1] "A" "B" "C"

[[2]]
[1] "ab" "bc" "cd"

##### S-Plus produced the following:
 > Cut(a2, 2)
   ab bc cd
A  1  1  2
B  1  1  2
C  1  2  2
attr(, "levels"):
[1] "0.92+ thru 5.00" "5.00+ thru 9.08"

#####
hope this helps.
spencer graves

Peter Dalgaard BSA wrote:
> Tamas Papp <tpapp at axelero.hu> writes:
> 
> 
>>Dear list,
>>
>>I'd like to use the function cut() on matrices, ie that when I apply
>>it to a matrix, it would return a matrix of the same dimensions
>>instead of a vector.
>>
>>I wonder if there is a better (more elegant) solution than
>>
>>matrix(cut(a, ...), ncol=ncol(a), nrow=nrow(a))
>>
>>because I would like to use cut on both vectors and matrices and avoid
>>testing whether a is a matrix.
> 
> 
> Will this not work?:
> 
> ac <- cut(a)
> dim(ac) <- dim(a)
>