R Shuffle and randomize columns of a data table

user_7350570 0
Viewed 388

Given a data.table such as:

data.table::data.table(a = c(1,2,3), b = c("red","blue","yellow"), c = c(TRUE, FALSE, TRUE), d = c(21, 45, 34, 26))

   a      b     c  d
1: 1    red  TRUE 21
2: 2   blue FALSE 45
3: 3 yellow  TRUE 34
4: 4  green FALSE 26

where a is a unique row identifier, how could I randomize/anonymize the data so that the columns shuffle within their own column. This would create a random data.table that looks something like:

   a      b     c  d
1: 1  green  TRUE 26
2: 2 yellow FALSE 45
3: 3    red FALSE 21
4: 4   blue  TRUE 34
data.table r
1 Answers

If it is random for each column that doesn't tie with the rows, then use sample on the columns specified in the .SDcols by looping over them in lapply and assign (:=) the output back to the columns

dt1[, (2:4) := lapply(.SD, sample), .SDcols = 2:4]

-output

dt1
#   a      b     c  d
#1: 1   blue FALSE 34
#2: 2    red  TRUE 21
#3: 3  green FALSE 45
#4: 4 yellow  TRUE 26

Or another option is set

for(j in names(dt1)[-1]) {
      set(dt1, i = NULL, j = j, value = sample(dt1[[j]]))
 }

data

dt1 <- data.table::data.table(
      a = c(1,2,3, 4),
      b = c("red","blue","yellow", "green"), 
      c = c(TRUE, FALSE, TRUE, FALSE), 
      d = c(21, 45, 34, 26)
   )
user_37322710
Related