Question

I have a input vector vi with boolean values. I want to take a random sample of size n from the vector where the value is true, so the final vector vf has these properties

1. The lengths of the vectors are equal
   length(vf) == length(v0)

2. vf has n true values
   n==sum(vf)

3. The true values in vf cannot be more than those in v0
   n <= sum(v0)

4. All the true values in vf are also true in vi

The vectors represents a selection of rows in a data frame, and this implements a stratified sample. So far I figured out how to use which() to get the row numbers, to use sample() to get a random sample, but the last part is recreating the boolean vector. There is probably a more elegant way?

For example:

n <- 1

v0 <- c(T,T,F,F)

vf <- c(T,F,F,F)

Answer 1

Here’s one solution:

# Make up some vector v0 and choose n
v0 <- rep(c(F,T,F), 5)
n <- 3


# The actual code
x <- which(v0)
vf <- logical(length(v0))
vf[x[sample.int(length(x), n)]] <- TRUE


# Finally validate the result
identical(length(vf), length(v0)) # TRUE
all(v0[vf])  # TRUE
sum(vf) == n # TRUE