IE381 - Calculating GINI and Entropy

gistfile1.py

import math 


def cal_entropy(lst):	
	t = sum(lst)
	result = 0


	for i in lst:
		division = float(i)/float(t)
#		print  "-", i, "/", t, "*", "math.log(", i, "/", t, ",", 2,")"
		result = result - division*math.log(division,2)
		
#	print "=", result
	return result  


def cal_gini(lst):
	t = sum(lst)
	result = 1.0

	for i in lst:
		result = result - (float(i)/float(t))**2
	
	return result


def entropy_average(attr1, attr2):
	result1 = cal_entropy(attr1)
	result2 = cal_entropy(attr2)
	
	total = float(sum(attr1)+sum(attr2))
	
#	print "1. ",attr1, "->", result1
#	print "2. ",attr2, "->", result2	
	

	sum1 = (sum(attr1)/total)*result1 + (sum(attr2)/total)*result2

	return sum1 
	
def gini_average(attr1, attr2):
	result1 = cal_gini(attr1)
	result2 = cal_gini(attr2)
	
	total = float(sum(attr1)+sum(attr2))
	
#		print "1. ",attr1, "->", result1
#		print "2. ",attr2, "->", result2	
	

	sum1 = (sum(attr1)/total)*result1 + (sum(attr2)/total)*result2

	return sum1 

no = [2,3]
yes = [2,3,4]

print "no:", cal_gini(no) , "yes:", cal_gini(yes)
print "gini:",gini_average(no, yes)
print "-"
print "no:", cal_entropy(no), "yes:", cal_entropy(yes)
print "entropy:", entropy_average(no, yes)