def test_3D_hierarchical_clusters():
# Simple three dimensional test
cluster_1 = ('c1', 0, 0, 0)
cluster_2 = ('c2', 10, 10, 10)
# Create a set of 54 3D items positioned in two separate clusters
SCATTER = (-1, 0, 1)
items = []
for i in SCATTER:
for j in SCATTER:
for k in SCATTER:
for c_name, c_x, c_y, c_z in (cluster_1, cluster_2):
name = '%s_(%d,%d,%d)' % (c_name, i, j, k)
x, y, z = (c_x + i), (c_y + j), (c_z + k)
items.append(Item(name, (x, y, z)))
n_items = len(items)
clusters = find_clusters(items, ("c1", "c2"))
# We did not loose, or add any item
c_items = sum(map(len, [c.items for c in clusters]))
assert n_items == c_items, "Error n items: %d" % (n_items - c_items)
# Items belongs to the correct cluster
for c in clusters:
name = c.items[0].name[:2]
for i in c.items:
assert i.name.startswith(name), 'Error: "%s" != "%s"' % (i.name[:2], name)
def test_langpop_hierarchical_clusters():
clusters = find_clusters([Item(lang, n) for lang, n in LANGUAGES_POPULARITY],
('Ubiquitous', 'Very Popular', 'Popular', 'Niche'))
for cluster, (expected_label, expected_list) in zip(clusters, EXPECTED_CLUSTERS):
assert cluster.name == expected_label
cluster.items.sort(reverse=True)
assert [item.name for item in cluster.items] == expected_list
