How do I create component (subset) dataframes in R based on column values?

I'd like to split a dataframe into several component dataframes based on the values in one column. In my example, I want to split dat into dat.1, dat.2 and dat.3 using the values in column "cond". Is there a simple command which could achieve this?

dat
sub cond   开发者_如何学Python trial   time01  time02
1   1   1   2774    8845
1   1   2   2697    9945
1   2   1   2219    9291
1   2   2   3886    7890
1   3   1   4011    9032
2   2   1   3478    8827
2   2   2   2263    8321
2   3   1   4312    7576
3   1   1   4219    7891
3   3   1   3992    6674


dat.1               
sub cond    trial   time01  time02
1   1   1   2774    8845
1   1   2   2697    9945
3   1   1   4219    7891    

dat.2               
sub cond    trial   time01  time02
2   2   1   3478    8827
2   2   2   2263    8321
1   2   1   2219    9291
1   2   2   3886    7890

dat.3               
sub cond    trial   time01  time02
1   3   1   4011    9032
2   3   1   4312    7576
3   3   1   3992    6674

Perhaps because I'm an R novice I've still not determined how to do this despite browsing and trying the solutions proposed in several similar forum queries. Thank you in advance for any replies.

A dput() of the data is:

structure(list(sub = c(1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 3L, 3L
), cond = c(1L, 1L, 2L, 2L, 3L, 2L, 2L, 3L, 1L, 3L), trial = c(1L, 
2L, 1L, 2L, 1L, 1L, 2L, 1L, 1L, 1L), time01 = c(2774L, 2697L, 
2219L, 3886L, 4011L, 3478L, 2263L, 4312L, 4219L, 3992L), time02 = c(8845L, 
9945L, 9291L, 7890L, 9032L, 8827L, 8321L, 7576L, 7891L, 6674L
)), .Names = c("sub", "cond", "trial", "time01", "time02"), class = "data.frame", row.names = c(NA, 
-10L))

I think the easiest way is via split:

split(dat, dat$cond)

Note however, that split returns a list of the data.frames.

To obtain single data.frames from the list you could procede as follows using a loop to make the single objects (implicit in the lapply statement):

tmp <- split(dat, dat$cond)
lapply(1:length(tmp), function(x) assign(paste("dat.", x, sep = ""), tmp[[x]], envir = .GlobalEnv))

However, using a list is probably more Rish and will be more useful in the long run.

Thanks to Gavin for posting the data!

Is there anything not satisfying about

split(dat, dat$cond)

? You do have R and split as tags, you know...

Yes, split(). For example, if your data are in dat, then:

with(dat, split(dat, cond))

returns a list, whose components are the data frames you wanted:

R> with(dat, split(dat, cond))
$`1`
  sub cond trial time01 time02
1   1    1     1   2774   8845
2   1    1     2   2697   9945
9   3    1     1   4219   7891

$`2`
  sub cond trial time01 time02
3   1    2     1   2219   9291
4   1    2     2   3886   7890
6   2    2     1   3478   8827
7   2    2     2   2263   8321

$`3`
   sub cond trial time01 time02
5    1    3     1   4011   9032
8    2    3     1   4312   7576
10   3    3     1   3992   6674

Just for the sake of completeness, here's a way to do it with the plyr package.

require(plyr)

> dlply( dat, .(cond))
$`1`
  sub cond trial time01 time02
1   1    1     1   2774   8845
2   1    1     2   2697   9945
9   3    1     1   4219   7891

$`2`
  sub cond trial time01 time02
3   1    2     1   2219   9291
4   1    2     2   3886   7890
6   2    2     1   3478   8827
7   2    2     2   2263   8321

$`3`
   sub cond trial time01 time02
5    1    3     1   4011   9032
8    2    3     1   4312   7576
10   3    3     1   3992   6674

attr(,"class")
[1] "split" "list" 

Note the syntactic simplicity in that you only mention dat once.

;)

ucond <- unique(dat$cond)
dat_by_cond <- lapply(lapply(ucond, "==", dat$cond), subset, x=dat)
names(dat_by_cond) <- paste("dat",ucond,sep=".")