8

I'm constructing a data.table from two (or more) input vectors with different lengths:

x <- c(1,2,3,4)
y <- c(8,9)

dt <- data.table(x = x, y = y)

And need the shorter vector(s) to be filled with NA rather than recycling their values, resulting in a data.table like this:

   x  y
1: 1  8
2: 2  9
3: 3 NA
4: 4 NA

Is there a way to achieve this without explicitly filling the shorter vector(s) with NA before passing them to the data.table() constructor?

Thanks!

Matheus Lacerda
  • 5,983
  • 11
  • 29
  • 45
Steffen J.
  • 694
  • 5
  • 16

3 Answers3

8

One can use out of range indices:

library("data.table")

x <- c(1,2,3,4)
y <- c(8,9)
n <- max(length(x), length(y))

dt <- data.table(x = x[1:n], y = y[1:n])
# > dt
#    x  y
# 1: 1  8
# 2: 2  9
# 3: 3 NA
# 4: 4 NA

Or you can extend y by doing (as @Roland recommended in the comment):

length(y) <- length(x) <- max(length(x), length(y))
dt <- data.table(x, y)
Frank
  • 66,179
  • 8
  • 96
  • 180
jogo
  • 12,469
  • 11
  • 37
  • 42
3

An option is cbind.fill from rowr

library(rowr)
setNames(cbind.fill(x, y, fill = NA), c("x", "y"))

Or place the vectors in a list and then pad NA at the end based on the maximum length of the list elements

library(data.table)
lst <- list(x = x, y = y)
as.data.table(lapply(lst, `length<-`, max(lengths(lst))))
#   x  y
#1: 1  8
#2: 2  9
#3: 3 NA
#4: 4 NA
akrun
  • 874,273
  • 37
  • 540
  • 662
3

The "out of range indices" answer provided by jogo can be extended cleanly to in-place assignment using .N:

x <- c(1,2,3,4)
y <- c(8,9)
n <- max(length(x), length(y))
dt <- data.table(x = x[1:n], y = y[1:n])

z <- c(6,7)
dt[, z := z[1:.N]]
#    x  y  z
# 1: 1  8  6
# 2: 2  9  7
# 3: 3 NA NA
# 4: 4 NA NA
arau
  • 61
  • 2