Question

I’m a bit at a loss as to how to find a clean algorithm for doing the following:

Suppose I have a dict k:

>>> k = {'A': 68, 'B': 62, 'C': 47, 'D': 16, 'E': 81}

I now want to randomly select one of these keys, based on the ‘weight’ they have in the total (i.e. sum) amount of keys.

>>> sum(k.values()) 
>>> 274

So that there’s a

>>> 68.0/274.0
>>> 0.24817518248175183

24.81% percent change that A is selected.

How would you write an algorithm that takes care of this? In other words, that makes sure that on 10.000 random picks, A will be selected 2.481 times?

Answer 1

Here’s a weighted choice function, with some code that exercises it.

import random

def WeightedPick(d):
    r = random.uniform(0, sum(d.itervalues()))
    s = 0.0
    for k, w in d.iteritems():
        s += w
        if r < s: return k
    return k

def Test():
    k = {'A': 68, 'B': 62, 'C': 47, 'D': 16, 'E': 81}
    results = {}
    for x in xrange(10000):
        p = WeightedPick(k)
        results[p] = results.get(p, 0) + 1
    print results

Test()