def test_sample_weight_consistency():
# given
dbscan = DBScanStream(eps=1.2, min_samples=5)
X = np.array([[1, 1]])
sample_weight = np.array([1, -1, 3, 1, 2])
# when (sample weights shape is not consistent with input data)
with pytest.raises(AssertionError):
dbscan.fit(X, sample_weight)
def test_outliers_only():
# given
dbscan = DBScanStream(eps=1, min_samples=5)
X = np.array([[1, 2],
[2, 2]])
# when
clustering = dbscan.fit(X)
# then (outliers label is -1)
assert np.array_equal(clustering.labels_, np.array([-1, -1]))
def test_one_cluster():
# given
dbscan = DBScanStream(eps=1.42, min_samples=3)
X = np.array([[1, 1],
[2, 2],
[0, 1]])
# when
clustering = dbscan.fit(X)
# then (outliers label is -1)
assert np.array_equal(clustering.labels_, np.array([0, 0, 0]))
def test_two_clusters_with_outliers():
# given
dbscan = DBScanStream(eps=1.42, min_samples=2)
X = np.array([[1, 1],
[2, 2],
[5, 6],
[6, 7],
[-1, -2],
[56, 34]])
# when
clustering = dbscan.fit(X)
# then (outliers label is -1)
assert np.array_equal(clustering.labels_, np.array([0, 0, 1, 1, -1, -1]))
def test_negative_sample_weights():
# given
dbscan = DBScanStream(eps=1.42, min_samples=2)
X = np.array([[1, 1],
[2, 2],
[5, 6],
[6, 7],
[-1, -2],
[56, 34]])
sample_weight = np.array([1, 1, -1, 3, 1, 2])
# when
clustering = dbscan.fit(X, sample_weight)
# then (outliers label is -1)
assert np.array_equal(clustering.labels_, np.array([0, 0, 1, 1, -1, 2]))