def test_annotate_projection(self):
clusters, clusters_meta, eps = annotate_projection(
self.annotations,
self.data,
clustering_algorithm=DBSCAN,
labels_per_cluster=2,
eps=3)
self.assertLessEqual(int(Orange.__version__.split(".")[1]), 22)
def test_example_not_clustered(self):
self.data[-1] = [23, 23]
clusters, clusters_meta, eps = annotate_projection(
self.annotations,
self.data,
clustering_algorithm=DBSCAN,
eps=2,
min_samples=3)
self.assertLessEqual(int(Orange.__version__.split(".")[1]), 22)
def test_other_clustering(self):
clusters, clusters_meta, locs = annotate_projection(
self.annotations,
self.data,
clustering_algorithm=KMeans,
n_clusters=2,
labels_per_cluster=2)
self.assertEqual(2, len(set(clusters.X.flatten())))
self.assertEqual(3, len(clusters_meta["C1"]))
self.assertEqual(3, len(clusters_meta["C2"]))
self.assertEqual(2, len(clusters_meta["C1"][0]))
self.assertEqual(2, len(clusters_meta["C2"][0]))
def compute_clusters(embedding: Table, result: Result, state: TaskState):
if not result.scores.table or not embedding:
result.clusters.table = None
result.clusters.groups = None
else:
state.set_status("Finding clusters...")
kwargs = {}
if result.clusters.epsilon is not None:
kwargs["eps"] = result.clusters.epsilon
clusters = annotate_projection(result.scores.table, embedding, **kwargs)
result.clusters.table = clusters[0]
result.clusters.groups = clusters[1]
result.clusters.epsilon = clusters[2]
state.set_partial_result(("clusters", result))
def test_one_label(self):
"""
Test whether having only one label works fine, in this case one cluster
will not have a label assigned - this must work fine as well.
"""
domain_ann = Domain([ContinuousVariable("a")])
data_ann = np.array([[0.9], [0.9], [0.9], [0.9], [0.9], [0], [0], [0],
[0], [0], [0], [0]])
ann = Table(domain_ann, data_ann)
clusters, clusters_meta, eps = annotate_projection(ann,
self.data,
eps=2)
self.assertGreater(len(clusters), 0)
self.assertGreater(len(clusters_meta), 0)
self.assertGreater(len(clusters_meta["C1"][0]), 0)
self.assertGreater(len(clusters_meta["C1"][1]), 0)
self.assertGreater(len(clusters_meta["C1"][2]), 0)
def test_example_not_clustered(self):
self.data[-1] = [23, 23]
clusters, clusters_meta, eps = annotate_projection(
self.annotations,
self.data,
clustering_algorithm=DBSCAN,
eps=2,
min_samples=3)
self.assertListEqual(
list(map(clusters.domain.attributes[0].repr_val,
clusters.X[:, 0])), ["C2"] * 5 + ["C1"] * 6 + ["?"])
self.assertEqual(len(clusters_meta), 2)
self.assertEqual(len(clusters_meta["C1"]), 3)
self.assertEqual(len(clusters_meta["C2"]), 3)
self.assertEqual(clusters_meta["C1"][0][0][0], 'c')
self.assertAlmostEqual(clusters_meta["C1"][0][0][1], 0.5, 5)
self.assertEqual(clusters_meta["C1"][0][1][0], 'b')
self.assertAlmostEqual(clusters_meta["C1"][0][1][1], 0.5, 5)
def test_annotate_projection(self):
clusters, clusters_meta, eps = annotate_projection(
self.annotations,
self.data,
clustering_algorithm=DBSCAN,
labels_per_cluster=2,
eps=3.0)
self.assertListEqual(
list(map(clusters.domain.attributes[0].repr_val,
clusters.X[:, 0])), ["C2"] * 5 + ["C1"] * 7)
self.assertEqual(len(clusters_meta), 2)
self.assertEqual(len(clusters_meta["C1"]), 3)
self.assertEqual(len(clusters_meta["C2"]), 3)
self.assertEqual(clusters_meta["C1"][0][0][0], 'b')
self.assertAlmostEqual(clusters_meta["C1"][0][0][1], 4 / 7, 5)
self.assertEqual(clusters_meta["C1"][0][1][0], 'c')
self.assertAlmostEqual(clusters_meta["C1"][0][1][1], 3 / 7, 5)
self.assertEqual(2, len(clusters_meta["C1"][1]))
self.assertEqual(2, clusters_meta["C1"][2].shape[1])
self.assertEqual(2, clusters_meta["C1"][2].shape[1])
self.assertEqual(float, type(eps))
def test_empty_annotations(self):
ann = Table(Domain([]), np.empty((len(self.data), 0)))
clusters, clusters_meta, eps = annotate_projection(ann, self.data)
self.assertGreaterEqual(len(clusters), 0)
self.assertEqual(len(clusters_meta), 0)