How to partition a set of values (vector) in R

Question

I'm programming in R. I've got a vector containing, let's say, 1000 values. Now let's say I want to partition these 1000 values randomly into two new sets, one containing 400 values and the other containing 600. How could I do this? I've thought about doing something like this...

firstset <- sample(mydata, size=400)

...but this doesn't partition the data (in other words, I still don't know which 600 values to put in the other set). I also thought about looping from 1 to 400, randomly removing 1 value at a time and placing it in firstset. This would partition the data correctly, but how to implement this is not clear to me. Plus I've been told to avoid for loops in R whenever possible.

Any ideas?

Answer 1

Instead of sampling the values, you could sample their positions.

positions <- sample(length(mydata), size=400)  # ucfagls' suggestion
firstset <- mydata[positions]
secondset <- mydata[-positions]

EDIT: ucfagls' suggestion will be more efficient (especially for larger vectors), since it avoids allocating a vector of positions in R.

Answer 2

Just randomize mydata and take the first 400 and then last 600.

mydata <- sample(mydata)
firstset <- mydata[1:400]
secondset <- mydata[401:1000]

Answer 3

If mydata is truly a vector, one option would be:

split(mydata, sample(c(rep("group1", 600), rep("group2", 400))))