Group by ID, each element of the new table is a vector

Question

I have a table like this

data.table(ID = c(1,2,3,4,5,6), 
         R = c(\"s\",\"s\",\"n\",\"n\",\"s\",\"s\"), 
         S = c(\"a\",\"a\",\"a\",\"b\",\"b\",         


        

Answer 1

You could try:

library(dplyr)
library(tidyr)

df %>% 
  group_by(R, S) %>% 
  summarise(i = toString(ID)) %>% 
  spread(S, i) 

Which gives:

#Source: local data table [2 x 3]
#Groups: 
#
#  R    a    b
#1 n    3    4
#2 s 1, 2 5, 6

Note: This will store the result in a string. If you want a more convenient format to access the elements, you could store in a list:

df2 <- df %>% 
  group_by(R, S) %>% 
  summarise(i = list(ID)) %>% 
  spread(S, i)  

Which gives:

#Source: local data table [2 x 3]
#Groups: 
#
#  R        a        b
#1 n <dbl[1]> <dbl[1]>
#2 s <dbl[2]> <dbl[2]>

You can then access the elements by doing:

> df2$a[[2]][2]
#[1] "2"

Answer 2

You can use dcast from reshape2 with the appropriate aggregating function:

library(functional)
library(reshape2)

dcast(df, R~S, value.var='ID', fun.aggregate=Curry(paste0, collapse=','))
#  R   a   b
#1 n   3   4
#2 s 1,2 5,6

Or even short as @akrun underlined:

dcast(df, R~S, value.var='ID', toString)

Answer 3

Here's an alternative that uses plain old data.table syntax:

DT[,lapply(split(ID,S),list),by=R]

# or...

DT[,lapply(split(ID,S),toString),by=R]