Weighted random from int list

Question

I have a list of integers and I would like to be able to choose a random number based on the weights in a that list.

For example let's say I have a list like this: List = [20,40,80,60].

The chances of me choosing the number 1 (or 0 depending on how you see it) would be 10% and the chance of getting the number 3 would be 30%.

How would I go about doing this in a pythonic way?

Have you seen this? http://stackoverflow.com/questions/14992521/python-weighted-random — Casey Falk, Aug 06 '14 at 12:24

unutbu · Accepted Answer · 2014-08-06T17:36:24.407

A simple algorithm might look like this:

import bisect
import random
def random_choice(choices, weights):
    """
    >>> a = ['Hit', 'Out']
    >>> b = [.3, .7]
    >>> random_choice(a,b)
    """
    cumsums = []
    c = 0
    for weight in weights:
        c += weight
        cumsums.append(c)
    rnd = random.uniform(0, c)
    i = bisect.bisect(cumsums, rnd)
    return choices[i]

Demo:

>>> random_choice(list('ABCD'), weights=[20,40,80,60])
'C'

However, there are algorithms which are more efficient if you plan on calling random_choice many times. See for example, the Alias method and this article for a comparison of many options.

Thank you, it worked like a charm. Didn't exactly fit what I needed, but I made it work. — TumbaBit, Aug 06 '14 at 15:19

Weighted random from int list

1 Answers1