strsplit by row and distribute results by column in data.frame

Question

So I have the data.frame

dat = data.frame(x = c('Sir Lancelot the Brave', 'King Arthur',  
                       'The Black Knight', 'The Rabbit'), stringsAsFactors=F)

> dat
                       x
1 Sir Lancelot the Brave
2            King Arthur
3       The Black Knight
4             The Rabbit

And I want to transform it into the data frame

> dat2
                       x    1            2       3      4
1 Sir Lancelot the Brave    Sir   Lancelot     the  Brave
2            King Arthur    King    Arthur
3       The Black Knight    The      Black  Knight 
4             The Rabbit    The     Rabbit

strsplit returns the data as a list

sbt <- strsplit(dat$x, " ")
> sbt
[[1]]
[1] "Sir"      "Lancelot" "the"      "Brave"   

[[2]]
[1] "King"   "Arthur"

[[3]]
[1] "The"    "Black"  "Knight"

[[4]]
[1] "The"    "Rabbit"

and as.data.table does not create NULL values where it should, but repeats values

> t(as.data.table(sbt))
   [,1]   [,2]       [,3]     [,4]    
V1 "Sir"  "Lancelot" "the"    "Brave" 
V2 "King" "Arthur"   "King"   "Arthur"
V3 "The"  "Black"    "Knight" "The"   
V4 "The"  "Rabbit"   "The"    "Rabbit"

I guess I really would like an argument to as.data.table(x, repeat=FALSE), else how can I accomplish this job?

Answer 1

This is an old question, I know, but I thought I would share two additional options.

Option 1

concat.split from my "splitstackshape" package was designed exactly for this type of thing.

library(splitstackshape)
concat.split(dat, "x", " ")
#                        x  x_1      x_2    x_3   x_4
# 1 Sir Lancelot the Brave  Sir Lancelot    the Brave
# 2            King Arthur King   Arthur             
# 3       The Black Knight  The    Black Knight      
# 4             The Rabbit  The   Rabbit        

---

Option 2

data.table has recently (as of version 1.8.11, I believe) had some additions to its arsenal, notably in this case dcast.data.table. To use it, unlist the split data (as was done in @mnel's answer), create a "time" variable using .N (how many new values per row), and use dcast.data.table to transform the data into the form you are looking for.

library(data.table)
library(reshape2)
packageVersion("data.table")
# [1] ‘1.8.11’

DT <- data.table(dat)
S1 <- DT[, list(X = unlist(strsplit(x, " "))), by = seq_len(nrow(DT))]
S1[, Time := sequence(.N), by = seq_len]
dcast.data.table(S1, seq_len ~ Time, value.var="X")
#    seq_len    1        2      3     4
# 1:       1  Sir Lancelot    the Brave
# 2:       2 King   Arthur     NA    NA
# 3:       3  The    Black Knight    NA
# 4:       4  The   Rabbit     NA    NA

Answer 2

Here's one option. The single complication is that you need to first convert each vector to a data.frame with one row, as data.frames are what rbind.fill() expects.

library(plyr)
rbind.fill(lapply(sbt, function(X) data.frame(t(X))))
#     X1       X2     X3    X4
# 1  Sir Lancelot    the Brave
# 2 King   Arthur   <NA>  <NA>
# 3  The    Black Knight  <NA>
# 4  The   Rabbit   <NA>  <NA>

My own inclination, though, would be to just use base R, like this:

n <- max(sapply(sbt, length))
l <- lapply(sbt, function(X) c(X, rep(NA, n - length(X))))
data.frame(t(do.call(cbind, l)))
#     X1       X2     X3    X4
# 1  Sir Lancelot    the Brave
# 2 King   Arthur   <NA>  <NA>
# 3  The    Black Knight  <NA>
# 4  The   Rabbit   <NA>  <NA>

Answer 3

sbt = strsplit(dat$x, " ")
sbt
#[[1]]
#[1] "Sir"      "Lancelot" "the"      "Brave"   
#[[2]]
#[1] "King"   "Arthur"
#[[3]]
#[1] "The"    "Black"  "Knight"
#[[4]]
#[1] "The"    "Rabbit"

ncol = max(sapply(sbt,length))
ncol
# [1] 4

as.data.table(lapply(1:ncol,function(i)sapply(sbt,"[",i)))
#      V1       V2     V3    V4
# 1:  Sir Lancelot    the Brave
# 2: King   Arthur     NA    NA
# 3:  The    Black Knight    NA
# 4:  The   Rabbit     NA    NA

Answer 4

Using data.table as it appears you are trying to use it.

library(data.table)
DT <- data.table(dat)
DTB <- DT[, list(y = unlist(strsplit(x, ' '))), by = x]

new <- rep(NA_character_,  DTB[,.N,by =x][which.max(N), N])
names(new) <- paste0('V', seq_along(new))
DTB[,{.new <- new 
      .new[seq_len(.N)] <- y 
       as.list(.new)} ,by= x]

Or using reshape2 dcast to reshape

library(reshape2)

dcast(DTB[,list(id = seq_len(.N),y),by= x ], x ~id, value.var = 'y')

Answer 5

Here is a nice and simple approach with tidyr.

library(tidyr)

ncol <- max(sapply(dat, length))

dat %>%
  separate(x, paste0("V", seq(1,ncol)))

Note: You will get a warning, however, it is basically telling you that separate is padding the data with NA's. So you can ignore the warning.