Loyd's Algorithm for Kmeans

#!/usr/bin ipython
import numpy as np

# data
X = np.asarray([(1.0, 1.0), (1.5, 2.0), (3.0, 4.0), (5.0, 7.0), (3.5, 5.0), (4.5, 5.0), (3.5, 4.5)])

# Loyds algorithm with Forgy method
import random

# euclidian distance
def dist(x1, x2):
	return np.linalg.norm(x1-x2)

# Takes n means in n dimensions
def find_center(points):
	return reduce(np.add, points) / len(points)

def init(X, k=3, iterations = 10):
	centers = random.sample(X, k)
	
	while iterations > 0:
		print centers
		iterations -= 1

		# since numpy arrays aint hashable
		groups = { id(c): [] for c in centers }

		# assign points to groups
		for x in X:
			center = min(centers, key=lambda i: dist(i, x))
			groups[id(center)].append(x)

		centers = map(find_center, groups.itervalues())

	return centers

print init(X)