Probabilistic Chooser

Pick.py

"""Given a list of items, each with a floating-point number representing popularity, bounded at the bottom at zero (no negative values), pick an item based on the weight of that popularity.
"""

#NOTE: A variant of this is used in the Network flow class in pangaia.  <github/Social-Garden>\
#This was originally developed for genetic algorithms and the program called GA-SOLVE written in Pascal in 1992
# Zero-weighted values are given a value equal to *half* the lowest value.

from random import *

cityrank = {"Denver": 0.5, "Chicago": 1.0, "Houston": 0.0, "NYC": 2.9}

sumrank, ranksaccumulated = 0.0, {}

for city in cityrank:
    sumrank += cityrank[city]
    ranksaccumulated[city] =  sumrank
for city in city

#Now pick a city based on the ranking.
throw = uniform(0.0,sumrank) #XXX need to check if uniform is right function, particularly at end-point
for city in cityrank: 
    if throw < ranksaccumulated[city]:
       break

print city

#this is a crude mockup, untested to document the theory that actually was implemented in the program called GA-SOLVE.