How can I create a list of structures automatically with R?

Lets say that开发者_开发技巧 RES is a list with capacity for 1000 structs that function kmeans generates as output.

How do I declare RES?

After having RES declared I want to do something like this:

for (i in 1:1000) {
  RES[i] = kmeans(iris,i)
}

Thank you. Rui

If you use the R apply idiom, your code will be simpler and you don't have to declare your variable in advance:

RES <- lapply(1:3, function(i)kmeans(dist(iris[, -5]),i))

The results:

> str(RES)
List of 3
 $ :List of 7
  ..$ cluster     : Named int [1:150] 1 1 1 1 1 1 1 1 1 1 ...
  .. ..- attr(*, "names")= chr [1:150] "1" "2" "3" "4" ...
  ..$ centers     : num [1, 1:150] 2.89 2.93 3.04 2.96 2.93 ...
  .. ..- attr(*, "dimnames")=List of 2
  .. .. ..$ : chr "1"
  .. .. ..$ : chr [1:150] "1" "2" "3" "4" ...
  ..$ totss       : num 55479
  ..$ withinss    : num 55479
  ..$ tot.withinss: num 55479
  ..$ betweenss   : num 4.15e-10
  ..$ size        : int 150
  ..- attr(*, "class")= chr "kmeans"
 $ :List of 7
  ..$ cluster     : Named int [1:150] 1 1 1 1 1 1 1 1 1 1 ...
  .. ..- attr(*, "names")= chr [1:150] "1" "2" "3" "4" ...
  ..$ centers     : num [1:2, 1:150] 0.531 4.104 0.647 4.109 0.633 ...
  .. ..- attr(*, "dimnames")=List of 2
  .. .. ..$ : chr [1:2] "1" "2"
  .. .. ..$ : chr [1:150] "1" "2" "3" "4" ...
  ..$ totss       : num 55479
  ..$ withinss    : num [1:2] 863 9743
  ..$ tot.withinss: num 10606
  ..$ betweenss   : num 44873
  ..$ size        : int [1:2] 51 99
  ..- attr(*, "class")= chr "kmeans"
 $ :List of 7
  ..$ cluster     : Named int [1:150] 2 2 2 2 2 2 2 2 2 2 ...
  .. ..- attr(*, "names")= chr [1:150] "1" "2" "3" "4" ...
  ..$ centers     : num [1:3, 1:150] 3.464 0.5 5.095 3.438 0.622 ...
  .. ..- attr(*, "dimnames")=List of 2
  .. .. ..$ : chr [1:3] "1" "2" "3"
  .. .. ..$ : chr [1:150] "1" "2" "3" "4" ...
  ..$ totss       : num 55479
  ..$ withinss    : num [1:3] 2593 495 1745
  ..$ tot.withinss: num 4833
  ..$ betweenss   : num 50646
  ..$ size        : int [1:3] 62 50 38
  ..- attr(*, "class")= chr "kmeans"

I'll second that lapply is the right answer in this case. But there are many scenarios where a loop is necessary, and it's a good question.

R lists don't need to be declared as empty ahead of time, so the easiest way to do this would to just 'declare' RES as an empty list:

RES <- list()
for (i in 1:1000) {
   RES[i] = kmeans(iris,i)
}

R will just extend the list for each iteration.

Incidentally, this works even for non-sequential indexing:

newList <- list()
newList[5] <- 100

yields a list with slots 1 through 4 designed as NULL and the number 100 in the fifth slot.

This is all just to say that lists are very different beasts in R than the atomic vectors.

The way to create a list is unfortunately different from the usual way of creating a numeric vector.

# The "usual" way to create a numeric vector
myNumVec <- numeric(1000) # A numeric vector with 1000 zeroes...

# ...But there is also this way
myNumVec <- vector("numeric", 1000) # A numeric vector with 1000 zeroes...


# ...and that's the only way to create lists:

# Create a list with 1000 NULLs
RES <- vector("list", 1000)

So your example would become,

RES <- vector("list", 1000)
for (i in 1:1000) {
  RES[[i]] = kmeans(iris,i) 
}

(Note that kmeans doesn't like to be called with the iris data set directly like that though...)

But then again, lapply would do it too and in a more direct way as @Andrie showed.