CODEWITHSUNDEEP
r
anba
anba

Reputation: 583

R - subsetting data.frame in combinationar manner

I've data.frame like below:

Name Feature
A    1
B    2
C    4 
D    1
E    7 
F    5
G    2
H    2

I have to create from this data series of data.frames with three rows in commbinationar manner. In another words, I need to obtain

subsetted_data.frame_1

Name Feature
A    1
B    2
C    4

subsetted_data.frame_2

Name Feature
D    1
G    2
H    2

subsetted_data.frame_3

Name Feature
F    5
G    2
H    2

And so on - to finally create all possible combinations. I was tried to use split function (from data.table package), but it doesn't work. Whis is the easiest way to obtain this?

Upvotes: 1

Views: 45

Answers (1)

Rui Barradas
Rui Barradas

Reputation: 76402

You can use combn to get a matrix of indexes and then pass them as the argument to a lapply anonymous function.

cmb <- combn(nrow(dat), 3)

sub_data <- lapply(seq_len(ncol(cmb)), function(i) dat[cmb[, i], ])
names(sub_data) <- sprintf("subsetted_data.frame_%02d", seq_along(sub_data))

EDIT.

Following @AkselA's comment I have tried his code and, if run before setting the names like the code above does, the two results are the same in the sense of identical, meaning, they are exactly the same.

sub_data2 <- apply(cmb, 2, function(x) dat[x,])
identical(sub_data, sub_data2)
#[1] TRUE

DATA in dput format.

dat <-
structure(list(Name = structure(1:8, .Label = c("A", "B", "C", 
"D", "E", "F", "G", "H"), class = "factor"), Feature = c(1L, 
2L, 4L, 1L, 7L, 5L, 2L, 2L)), .Names = c("Name", "Feature"), class = "data.frame", row.names = c(NA, 
-8L))

Upvotes: 1

Related Questions