CODEWITHSUNDEEP
r
Peaceandlove
Peaceandlove

Reputation: 79

Random subsampling in column (r)

is possible to random subsample (i.e size 50) an entire column?

Input example:

Pa 0
Pb 0
Pc 127
Pd 0
Pe 13
Pf 39
Pg 0
Ph 113
Pi 0

Output example (size 50, random subsampled):

Pa 0
Pb 0
Pc 22
Pd 0
Pe 2
Pf 8
Pg 0
Ph 18
Pi 0

Any ideas?

Upvotes: 0

Views: 506

Answers (1)

akrun
akrun

Reputation: 887881

Try

indx <- df1$v2!=0
df1$v2[indx] <- sample(50, sum(indx), replace=FALSE)

Update

For getting the subsamples based on the condition that the values should be less than the original value

f1 <- function(x, n){
  indx <- x!=0
  v1 <- sample(n, sum(indx), replace=TRUE)
  while(any(v1 > x[indx])){
  v1 <- sample(n, sum(indx), replace=TRUE)
  }
 x[indx] <- v1
 x}

 set.seed(24)
 f1(df1$v2, 50)
 #[1]  0  0 15  0 12 36  0 26  0

Or use repeat

  f2 <- function(x, n){
   indx <- x!=0
   repeat{
    v1 <- sample(n, sum(indx), replace=TRUE)
    if(all(v1 <x[indx])) break
    }
   x[indx] <- v1
   x}
 set.seed(24)
 f2(df1$v2, 50)
 #[1]  0  0 15  0 12 36  0 26  0

data

df1 <- structure(list(v1 = c("Pa", "Pb", "Pc", "Pd", "Pe", "Pf", "Pg", 
"Ph", "Pi"), v2 = c(0L, 0L, 127L, 0L, 13L, 39L, 0L, 113L, 0L)), 
.Names = c("v1", 
"v2"), class = "data.frame", row.names = c(NA, -9L))

Upvotes: 1

Related Questions