def test_get_mean_1(self):
"""Test calculating mean on a single vector"""
vect = [1, 2, 3]
clus = Cluster()
clus.assign(vect)
np.testing.assert_array_equal(clus.get_mean(), vect)
def test_assign_retrieve_1(self):
"""Test retrieving a single vector"""
vect = [1, 2, 3]
clus = Cluster()
clus.assign(vect)
self.assertEqual(clus.get_samples(), [vect])
def test_get_distance_1(self):
"""Test Euclidean distance"""
vect = [1, 1, 1, 1]
clus = Cluster()
clus.assign(vect)
self.assertEqual(clus.get_distance([2, 2, 2, 2]), 2)
def test_assign_retrieve_n(self):
"""Test retrieving multiple vectors"""
v_0 = [1, 2, 3]
v_1 = [3, 2, 1]
v_2 = [4, 5, 6]
clus = Cluster()
clus.assign(v_0)
clus.assign(v_1)
clus.assign(v_2)
self.assertEqual(clus.get_samples(), [v_0, v_1, v_2])
def test_get_mean_n(self):
"""Test calculating mean on multiple vectors"""
v_0 = [2, 2, 2]
v_1 = [3, 5, 1]
v_2 = [4, 5, 6]
clus = Cluster()
clus.assign(v_0)
clus.assign(v_1)
clus.assign(v_2)
expected = [3, 4, 3]
np.testing.assert_array_equal(clus.get_mean(), expected)
def test_merge(self):
"""Test merging clusters"""
clus_a = Cluster()
clus_a.assign([1, 2, 3])
clus_a.assign([2, 3, 4])
clus_b = Cluster()
clus_b.assign([3, 4, 5])
clus_b.assign([4, 5, 6])
clus_a.merge(clus_b)
# test values (order not important)
self.assertCountEqual(clus_a.get_samples(),
[[1, 2, 3], [4, 5, 6], [2, 3, 4], [3, 4, 5]])
# test recalculated mean
np.testing.assert_array_equal(clus_a.get_mean(), [2.5, 3.5, 4.5])
def generate(data, num_clusters):
"""The common interface"""
clusters = []
# For each sample in the dataset
for index, sample in enumerate(data):
# Create initial clusters from first K samples
if index < num_clusters:
# print("Creating cluster:", index)
clust = Cluster()
clust.assign(sample)
clusters.append(clust)
continue
# "If the distance between the members of this pair..."
clusters = consolidate(clusters)
# "In addition, as each new point is processed..."
# Get distance from each cluster
distances = [c.get_distance(sample) for c in clusters]
if min(distances) > ROUGHENING:
# print("Adding a new cluster")
clust = Cluster()
clust.assign(sample)
clusters.append(clust)
# print("Num clusters is now:", len(clusters))
else:
# Assign to nearest
clusters[np.argmin(distances)].assign(sample)
clusters = consolidate(clusters)
return np.array([cl.get_mean() for cl in clusters])
