def test_radius_with_new_point():
micro_cluster = MicroCluster(0, 2, 0, 0.15)
point1 = np.array([0, 1])
radius = micro_cluster.radius_with_new_point(point1)
assert isinstance(radius, np.float64)
assert radius == 0
micro_cluster.cf1 = np.array([0, 0])
micro_cluster.cf2 = np.array([0, 1])
point1 = np.array([0, 1])
radius = micro_cluster.radius_with_new_point(point1)
assert radius == 1
micro_cluster.cf1 = np.array([0, 0])
micro_cluster.cf2 = np.array([1, 1])
point1 = np.array([0, 1])
radius = micro_cluster.radius_with_new_point(point1)
micro_cluster.cf1 = np.array([0, 1])
micro_cluster.cf2 = np.array([1, 2])
micro_cluster.weight = 1
assert radius == micro_cluster.radius
micro_cluster.cf1 = np.array([1, 1])
micro_cluster.cf2 = np.array([0, 0])
point1 = np.array([1, 1])
radius = micro_cluster.radius_with_new_point(point1)
micro_cluster.cf1 = np.array([2, 2])
micro_cluster.cf2 = np.array([1, 1])
micro_cluster.weight = 1
assert radius == 0
assert radius == micro_cluster.radius
def test_decay_micro_clusters(self, denstream):
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([5.0, 5.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 5
denstream.p_micro_clusters.append(micro_cluster)
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 5.0])
micro_cluster.cf2 = np.array([3.0, 0.0])
micro_cluster.weight = 10
denstream.p_micro_clusters.append(micro_cluster)
micro_cluster = MicroCluster(2, 2, 0, 0.15)
micro_cluster.cf1 = np.array([0.0, 0.0])
micro_cluster.cf2 = np.array([10.0, 2.0])
micro_cluster.weight = 3
denstream.p_micro_clusters.append(micro_cluster)
denstream.decay_micro_clusters(0)
assert np.all(denstream.p_micro_clusters[0].cf1 == np.array([5, 5]))
assert np.all(denstream.p_micro_clusters[0].cf2 == np.array([1, 1]))
assert np.all(denstream.p_micro_clusters[0].weight == 5)
assert np.all(denstream.p_micro_clusters[1].cf1 == np.array([1, 5]) *
2**(-0.15))
assert np.all(denstream.p_micro_clusters[1].cf2 == np.array([3, 0]) *
2**(-0.15))
assert np.all(denstream.p_micro_clusters[1].weight == 10 * 2**(-0.15))
assert np.all(denstream.p_micro_clusters[2].cf1 == np.array([0, 0]) *
2**(-0.15))
assert np.all(denstream.p_micro_clusters[2].cf2 == np.array([10, 2]) *
2**(-0.15))
assert np.all(denstream.p_micro_clusters[2].weight == 3 * 2**(-0.15))
def test_decay():
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.weight = 1
micro_cluster.cf1 = np.array([0.5, 0.7])
micro_cluster.cf2 = np.array([0.0, 1.0])
cf = micro_cluster.cf1.copy()
cf2 = micro_cluster.cf2.copy()
weight = micro_cluster.weight
micro_cluster.decay()
assert np.all(micro_cluster.cf1 == cf * (2**(-0.15)))
assert np.all(micro_cluster.cf2 == cf2 * (2**(-0.15)))
assert micro_cluster.weight == weight * (2**(-0.15))
def test_radius():
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.weight = 1
micro_cluster.cf1 = np.array([0, 0])
micro_cluster.cf2 = np.array([0, 1])
assert isinstance(micro_cluster.radius, np.float64)
assert micro_cluster.radius == 1.0
micro_cluster.cf1 = np.array([0, 0])
micro_cluster.cf2 = np.array([0, 2])
assert micro_cluster.radius == sqrt(2)
micro_cluster.cf1 = np.array([0, 1])
micro_cluster.cf2 = np.array([0, 2])
assert micro_cluster.radius == 1
micro_cluster.cf1 = np.array([0, 2])
micro_cluster.cf2 = np.array([0, 2])
assert micro_cluster.radius == 0
micro_cluster.cf1 = np.array([1, 2])
micro_cluster.cf2 = np.array([0, 0])
assert micro_cluster.radius == 0
def test_generate_clusters(self, denstream):
assert len(denstream.p_micro_clusters) == 0
dense_group, not_dense_group = denstream.generate_clusters()
assert dense_group == [[]]
assert not_dense_group == [[]]
case_0 = Case("0")
case_1 = Case("1")
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([5.0, 5.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 5
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
assert len(denstream.p_micro_clusters) == 1
assert micro_cluster.weight >= denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert not dense_group == [[]]
assert not_dense_group == [[]]
cluster_list = dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 5
assert cluster.case_ids == ["0", "1"]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([3.0, 3.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 3
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
assert len(denstream.p_micro_clusters) == 1
assert micro_cluster.weight < denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert dense_group == [[]]
assert not not_dense_group == [[]]
cluster_list = not_dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 3
assert cluster.case_ids == ["0", "1"]
case_2 = Case("2")
case_3 = Case("3")
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([3.0, 3.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 3
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([4.0, 4.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 4
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) <=
2 * denstream.epsilon)
assert cl1.weight + cl2.weight >= denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert not dense_group == [[]]
assert not_dense_group == [[]]
cluster_list = dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 3
assert cluster.case_ids == ["0", "1"]
cluster = cluster_list[1]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 4
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert dense_group[1]
assert cluster_list[2]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([3.0, 3.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 1
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([3.0, 3.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 1
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) <=
2 * denstream.epsilon)
assert cl1.weight + cl2.weight < denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert dense_group == [[]]
assert not not_dense_group == [[]]
cluster_list = not_dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [3, 3])
assert cluster.radius == 0
assert cluster.weight == 1
assert cluster.case_ids == ["0", "1"]
cluster = cluster_list[1]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [3, 3])
assert cluster.radius == 0
assert cluster.weight == 1
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert not_dense_group[1]
assert cluster_list[2]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([3.0, 3.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 3
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 1.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 4
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) >
2 * denstream.epsilon)
assert cl1.weight < denstream.mu
assert cl2.weight >= denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert not dense_group == [[]]
assert not not_dense_group == [[]]
cluster_list = not_dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 3
assert cluster.case_ids == ["0", "1"]
with pytest.raises(IndexError):
assert cluster_list[1]
cluster_list = dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [0.25, 0.25])
assert cluster.radius == cl2.radius
assert cluster.weight == 4
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert not_dense_group[1]
assert dense_group[1]
assert cluster_list[1]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([6.0, 6.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 6
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 1.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 2
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) >
2 * denstream.epsilon)
assert cl1.weight >= denstream.mu
assert cl2.weight < denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert not dense_group == [[]]
assert not not_dense_group == [[]]
cluster_list = dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 6
assert cluster.case_ids == ["0", "1"]
with pytest.raises(IndexError):
assert cluster_list[1]
cluster_list = not_dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [0.5, 0.5])
assert cluster.radius == cl2.radius
assert cluster.weight == 2
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert not_dense_group[1]
assert dense_group[1]
assert cluster_list[1]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([6.0, 6.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 6
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 1.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 4
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) >
2 * denstream.epsilon)
assert cl1.weight >= denstream.mu
assert cl2.weight >= denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert not dense_group == [[]]
assert not_dense_group == [[]]
cluster_list = dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [1, 1])
assert cluster.radius == 0
assert cluster.weight == 6
assert cluster.case_ids == ["0", "1"]
with pytest.raises(IndexError):
assert cluster_list[1]
cluster_list = dense_group[1]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [0.25, 0.25])
assert cluster.radius == cl2.radius
assert cluster.weight == 4
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert dense_group[2]
assert cluster_list[1]
micro_cluster = MicroCluster(0, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 1.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 3
denstream.p_micro_clusters = []
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_0.id] = micro_cluster.id
denstream.all_cases[case_1.id] = micro_cluster.id
micro_cluster = MicroCluster(1, 2, 0, 0.15)
micro_cluster.cf1 = np.array([1.0, 1.0])
micro_cluster.cf2 = np.array([1.0, 1.0])
micro_cluster.weight = 2
denstream.p_micro_clusters.append(micro_cluster)
denstream.all_cases[case_2.id] = micro_cluster.id
denstream.all_cases[case_3.id] = micro_cluster.id
cl1 = denstream.p_micro_clusters[0]
cl2 = denstream.p_micro_clusters[1]
assert len(denstream.p_micro_clusters) > 1
assert (denstream.euclidean_distance(cl1.centroid, cl2.centroid) >
2 * denstream.epsilon)
assert cl1.weight < denstream.mu
assert cl2.weight < denstream.mu
dense_group, not_dense_group = denstream.generate_clusters()
assert dense_group == [[]]
assert not not_dense_group == [[]]
cluster_list = not_dense_group[0]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 0
assert np.all(cluster.centroid == [(1 / 3), (1 / 3)])
assert cluster.radius == cl1.radius
assert cluster.weight == 3
assert cluster.case_ids == ["0", "1"]
with pytest.raises(IndexError):
assert cluster_list[1]
cluster_list = not_dense_group[1]
cluster = cluster_list[0]
assert isinstance(cluster, Cluster)
assert cluster.id == 1
assert np.all(cluster.centroid == [0.5, 0.5])
assert cluster.radius == cl2.radius
assert cluster.weight == 2
assert cluster.case_ids == ["2", "3"]
with pytest.raises(IndexError):
assert not_dense_group[2]
assert cluster_list[1]
def test_add_point(self, denstream: DenStream):
case_list = []
case_1 = Case("1")
case_1.add_event({
"concept:name": "activityA",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 00),
})
case_1.add_event({
"concept:name": "activityB",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 10),
})
case_1.add_event({
"concept:name": "activityC",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 20),
})
case_list.append(case_1)
case_2 = Case("2")
case_2.add_event({
"concept:name": "activityA",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 00),
})
case_2.add_event({
"concept:name": "activityB",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 10),
})
case_list.append(case_2)
graph = initialize_graph(case_list)
case_3 = Case("3")
case_3.add_event({
"concept:name": "activityA",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 00),
})
case_3.add_event({
"concept:name": "activityB",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 10),
})
case_3.add_event({
"concept:name": "activityC",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 20),
})
case_3.add_event({
"concept:name": "activityD",
"time:timestamp": datetime(2015, 5, 10, 8, 00, 30),
})
case_3.distances = calculate_case_distances(graph, case_3)
micro_cluster = MicroCluster(10, 2, 0, 0.15)
micro_cluster.cf1 = np.array([0.5, -0.5])
micro_cluster.cf2 = np.array([0.5, -0.1])
micro_cluster.weight = 10
denstream.p_micro_clusters.append(micro_cluster)
micro_cluster = MicroCluster(11, 2, 0, 0.15)
micro_cluster.cf1 = np.array([0.0, 0.0])
micro_cluster.cf2 = np.array([0.0, 0.0])
micro_cluster.weight = 5
denstream.o_micro_clusters.append(micro_cluster)
denstream.mc_id = 2
mc_id = denstream.add_point(case_3)
assert mc_id == 2
assert len(denstream.o_micro_clusters) == 2
assert denstream.o_micro_clusters[1].radius == 0
assert denstream.o_micro_clusters[1].weight == 1
assert np.all(denstream.o_micro_clusters[1].cf1 == case_3.point)
assert np.all(denstream.o_micro_clusters[1].cf2 == case_3.point *
case_3.point)
cf = denstream.o_micro_clusters[1].cf1.copy()
cf2 = denstream.o_micro_clusters[1].cf2.copy()
mc_id = denstream.add_point(case_3)
assert mc_id == 2
assert len(denstream.o_micro_clusters) == 1
assert len(denstream.p_micro_clusters) == 2
assert denstream.p_micro_clusters[1].weight == 2
assert np.all(denstream.p_micro_clusters[1].cf1 == cf + case_3.point)
assert np.all(denstream.p_micro_clusters[1].cf2 == cf2 +
case_3.point * case_3.point)