def k_means_clustering (k, centroids, points):
iterations = 0
while (True):
recalculatedCentroids = []
iterations += 1
# This is a 2-d list where each inner list is a list of points
# associated with the corresponding centroid/cluster
clusters = [[]] * k
for point in points:
new_clusterPoints = point.find_closest_point(centroids)
clusters[new_clusterPoints].append(point)
# recalculate the new centroids
for cluster in clusters:
recalculatedCentroids.append(Point.getAverage(cluster))
# check if centroids match
if (set(centroids) == set(recalculatedCentroids)):
print "Clustering completed in " + str(iterations) + " iterations"
return recalculatedCentroids
# re-iterate if new and old centroids do not match
centroids = copy.deepcopy(recalculatedCentroids)
