Question

Problem

I have N items of various types evenly distributed into their own buckets determined by type. I want to create a new list that:

1. randomly picks from each bucket
2. does not pick from the same bucket twice in a row
3. each bucket must have (if possible) an equal amount of representation in the final list
4. not using language specific libraries (not easily implemented in another language)

Example

I have 12 items of 4 distinct types which means I have 4 buckets:

Bucket A - [a, a, a]
Bucket B - [b, b, b]
Bucket C - [c, c, c]
Bucket D - [d, d, d]

What I want

A list of the above items in a random distribution without any characters repeating with a size between 1 and N.

12 Items: a, d, c, a, b, a, c, d, c, b, d, b

8 Items: c, a, d, a, b, d, c, b

4 Items: c, b, d, a

3 Items: b, c, a (Skipping D)

---

I was trying to do this with a while loop that generates random integers until the next bucket isn’t equal to the previously used bucket, but that seems inefficient, and was hoping someone else might have a better algorithm to solve this problem.

Answer 1

You could generate a random list of the buckets, and then randomly pick from then in order, removing the bucket from the list when you pick from it. When the list is empty, regenerate a random list of buckets, repeating until you pick the desired number of items.

Can you repeat items from the buckets? So if you pick the 1st “a” from bucket A the first time around, can you pick it a 2nd time? That’ll change the solution.