Subsetting a data frame by a value of one of its colums - r

Question

I have a rather large data frame. Here is a simplified example:
Group Element Value Note
1 AAA 11 Good
1 ABA 12 Good
1 AVA 13 Good
2 CBA 14 Good
2 FDA 14 Good
3 JHA 16 Good
3 AHF 16 Good
3 AKF 17 Good
Here it is as a dput:
dat <- structure(list(Group = c(1L, 1L, 1L, 2L, 2L, 3L, 3L, 3L), Element = structure(c(1L,
2L, 5L, 6L, 7L, 8L, 3L, 4L), .Label = c("AAA", "ABA", "AHF",
"AKF", "AVA", "CBA", "FDA", "JHA"), class = "factor"), Value = c(11L,
12L, 13L, 14L, 14L, 16L, 16L, 17L), Note = structure(c(1L, 1L,
1L, 1L, 1L, 1L, 1L, 1L), .Label = "Good", class = "factor")), .Names = c("Group",
"Element", "Value", "Note"), class = "data.frame", row.names = c(NA,
-8L))
I'm trying to separate it based on the group. so let's say
Group 1 will be a data frame:
Group Element Value Note
1 AAA 11 Good
1 ABA 12 Good
1 AVA 13 Good
Group 2:
2 CBA 14 Good
2 FDA 14 Good
and so on.

Answer 1

You can use split for this.
> dat
## Group Element Value Note
## 1 1 AAA 11 Good
## 2 1 ABA 12 Good
## 3 1 AVA 13 Good
## 4 2 CBA 14 Good
## 5 2 FDA 14 Good
## 6 3 JHA 16 Good
## 7 3 AHF 16 Good
## 8 3 AKF 17 Good
> x <- split(dat, dat$Group)
Then you can access each individual data frame by group number with x[[1]], x[[2]], etc.
For example, here is group 2:
> x[[2]] ## or x[2]
## Group Element Value Note
## 4 2 CBA 14 Good
## 5 2 FDA 14 Good
ADD: Since you asked about it in the comments, you can write each individual data frame to file with write.csv and lapply. The invisible wrapper is simply to suppress the output of lapply
> invisible(lapply(seq(x), function(i){
write.csv(x[[i]], file = paste0(i, ".csv"), row.names = FALSE)
}))
We can see that the files were created by looking at list.files
> list.files(pattern = "^[0-9].csv")
## [1] "1.csv" "2.csv" "3.csv"
And we can see the data frame of the third group with read.csv
> read.csv("3.csv")
## Group Element Value Note
## 1 3 JHA 16 Good
## 2 3 AHF 16 Good
## 3 3 AKF 17 Good

Answer 2

Obligatory plyr version (pretty much equiv to Richard's, but I'll bet it's slower, too:
library(plyr)
groups <- dlply(dat, .(Group), function(x) { return(x) })
length(groups)
## [1] 3
groups$`1` # can also do groups[[1]]
## Group Element Value Note
## 1 1 AAA 11 Good
## 2 1 ABA 12 Good
## 3 1 AVA 13 Good
groups[[2]]
## Group Element Value Note
## 1 2 CBA 14 Good
## 2 2 FDA 14 Good