Feature #12775: Random subset of array

• Status: Feedback
• Priority: Normal
• Assignee: -
• Target version: -

[sawa (Tsuyoshi Sawada)]

I often see use cases to randomly sample a subset from an array like this:

a = [3, 2, 6, 2, 6, 1]
a.sample(rand(a.length + 1)) # => [2, 6, 6, 3]

I request extending Array#sample to let it take an option (such as :arbitrary) for doing that:

a.sample(:arbitrary) # => [2, 6, 6, 3]

or perhaps having an independent method (like samples) to do it.

a.samples # => [2, 6, 6, 3]

[nobu (Nobuyoshi Nakada)]

• Status changed from Open to Feedback

Do you mean random size subset?
I can't get how/why it is useful.
Please elaborate with the use cases.

[sawa (Tsuyoshi Sawada)]

Nobuyoshi Nakada wrote:

Do you mean random size subset?
I can't get how/why it is useful.
Please elaborate with the use cases.

Yes, I mean random subset with random size. The use cases are mainly for testing.

[trans (Thomas Sawyer)]

This definition of #sample seems a bit limited. I know it aligns with the statical definition but it is very easy do another way: shuffle.take(n). Also, the interface is a little odd because it can return an element or an Array. Maybe it would be better if it could return repeats.

[1,2,3,4].sample(10) => [1,2,4,3,3,4,2,1,2,3]

Then #sample could also take a block and return an enumerator when given no argument. To get one element it would be sample.first.

But if backward compatibility needs to be preserved, then add this functionality by another name.

P.S. It would be weird but #sample could use negative numbers to allow repeating.

[shyouhei (Shyouhei Urabe)]

We looked at this issue at todays developer meeting but a "random subset with random size" still does not sound familiar to the attendees.

Can you show us in code how do you use such array?