How to select n unique elements at random, from a list of elements in Julia

Question

I want to generate n unique elements from a list of numbers. I came across this answer but that only gives me one element. I want n distinct elements from the list.

How do I go about doing this?

I have tried using rand(list,n) but this sometimes gives me repeated elements from list so that doesn't work.

Is the list small and in memory? If yes, you can shuffle it and take a prefix of length `n`. — phipsgabler, Oct 19 '21 at 06:39
Sometimes it is a list of numbers of variable length. Other times, it is a column of numbers from a `DataFrame` — newtothis, Oct 19 '21 at 06:44

DNF · Accepted Answer · 2021-10-19T13:36:37.150

13

Try ~~Distributions.sample~~ StatsBase.sample:

jl> using StatsBase: sample

jl> x = rand(10);

jl> sample(x, 3; replace=false)
3-element Vector{Float64}:
 0.6816165016249632
 0.8500982926818003
 0.6518188633423712

edited Oct 19 '21 at 13:36

answered Oct 19 '21 at 07:06

DNF

11,584
1
26
40

1

It looks like Distributions is just [reexporting sample from StatsBase](https://github.com/JuliaStats/Distributions.jl/blob/fb9ce8a1a3a3c860128c62bba8336140fe7dbef8/src/Distributions.jl#L262), but is there some benefit to using it via `Distributions` instead of directly using `StatsBase`? – Sundar R Oct 19 '21 at 12:28
I don't think so. I just cast about to see where `sample` was located, didn't find it in Statistics, so thought of Distributions.jl. I though Statistics depended on StasBase, so it never occurred to me to look there. – DNF Oct 19 '21 at 13:11

How to select n unique elements at random, from a list of elements in Julia

1 Answers1