def test_DAG_from_amat():
unconnected_amat = np.zeros((4, 4))
unconnected_graph = DAG.from_amat(unconnected_amat, list("ABCD"))
unconnected_graph_list = DAG.from_amat(unconnected_amat.tolist(),
list("ABCD"))
fully_connected_amat = np.tril(np.ones((4, 4)), -1)
fully_connected_graph = DAG.from_amat(fully_connected_amat, list("ABCD"))
assert DAG.from_amat(unconnected_amat).nodes == {"A", "B", "C", "D"}
assert np.all(unconnected_graph.get_numpy_adjacency() == unconnected_amat)
assert np.all(
unconnected_graph_list.get_numpy_adjacency() == unconnected_amat)
assert np.all(
fully_connected_graph.get_numpy_adjacency() == fully_connected_amat)
assert fully_connected_graph.nodes == unconnected_graph.nodes == {
"A", "B", "C", "D"
}
assert unconnected_graph.edges == set()
assert (fully_connected_graph.edges == fully_connected_graph.directed_edges
== {
('C', 'A'),
('B', 'A'),
('D', 'B'),
('D', 'C'),
('D', 'A'),
('C', 'B'),
})
def test_icdag_degree_caps():
dag = DAG.generate("ide_cozman", 10, burn_in=1000, max_indegree=2, seed=1)
assert max(dag.indegree()) <= 2
assert 2 < max(dag.outdegree(
)) <= 3, "Make sure out/degree isn't capped when indegree is set"
dag = DAG.generate("ide_cozman", 10, burn_in=1000, max_outdegree=2, seed=1)
assert max(dag.outdegree()) <= 2
assert 2 < max(dag.indegree(
)) <= 3, "Make sure in/degree isn't capped when outdegree is set"
dag = DAG.generate("ide_cozman", 10, burn_in=1000, max_degree=2, seed=1)
assert max(dag.degree()) <= 2
def test_compare(temp_out, test_dag):
img_path = temp_out / 'comparison.png'
dag2 = DAG()
dag2.add_vertices(list("ABCD"))
comparison_1 = test_dag.compare(dag2)
assert comparison_1.es['color'] == ['red'] * 3
assert comparison_1.es['style'] == ['dashed'] * 3
comparison_2 = dag2.compare(test_dag)
assert comparison_2.es['color'] == ['blue'] * 3
assert comparison_2.es['style'] == ['solid'] * 3
comparison_1.plot(img_path)
assert img_path.exists()
def test_DAG_serialise_continuous_str(test_dag):
dag = test_dag
dag.generate_continuous_parameters()
dag_string = dag.save()
dag2 = DAG.load(dag_string)
assert dag.nodes == dag2.nodes
assert dag.edges == dag2.edges
def test_DAG_serialise_discrete_str(test_dag):
dag = test_dag
dag.generate_discrete_parameters(seed=0)
dag_string = dag.save()
dag2 = DAG.load(dag_string)
assert dag.nodes == dag2.nodes
assert dag.edges == dag2.edges
def test_structure_types(structure_type):
dag_1 = DAG.generate(structure_type, 10, seed=1)
dag_2 = getattr(structure_generation,
structure_type.lower().replace(" ", "_"))(10, seed=1)
assert dag_1.is_dag()
assert dag_1.nodes == dag_2.nodes == set(ascii_uppercase[:10])
assert dag_1.edges == dag_2.edges > set()
def test_DAG_from_other():
test_graph = nx.DiGraph()
test_graph.add_nodes_from(list("ABCD"))
edges = [("C", "B"), ("D", "B"), ("D", "C")]
test_graph.add_edges_from(edges)
graph = DAG.from_other(test_graph)
assert graph.edges == graph.directed_edges == set(edges)
assert graph.nodes == set(list("ABCD"))
def test_DAG_serialise_discrete_file(temp_out, test_dag):
dag_path = temp_out / 'cont.pb'
dag = test_dag
dag.generate_discrete_parameters(seed=0)
dag.save(dag_path)
dag2 = DAG.load(dag_path)
assert dag.nodes == dag2.nodes
assert dag.edges == dag2.edges
def test_copy(test_dag):
dag = test_dag
dag_copy = dag.copy()
assert dag.nodes == dag_copy.nodes
assert dag.edges == dag_copy.edges
dag = DAG.generate("forest_fire", 10, seed=1)
dag_copy = dag.copy()
assert dag.nodes == dag_copy.nodes
assert dag.edges == dag_copy.edges
def test_bif_library():
all_bifs = [
p.stem for p in (Path(__file__).parent.parent / 'baynet' / 'utils' /
'bif_library').resolve().glob('*.bif')
]
for bif in all_bifs[:1]:
try:
dag = DAG.from_bif(bif)
data = dag.sample(1)
except Exception as e:
raise RuntimeError(f"Error loading {bif}: {e}")
def test_dfe_parameters():
dag = DAG.from_modelstring('[A]')
dag.vs['levels'] = [["A", "B"] for v in dag.vs]
data = pd.DataFrame({'A': [0, 0]})
dag.estimate_parameters(data,
method="dfe",
method_args={
"iterations": 1,
"learning_rate": 0.1
})
assert np.array_equal(dag.vs[0]['CPD'].cumsum_array, np.array([0.55, 1]))
def test_DAG_get_v_structures(test_dag, reversed_dag, partial_dag):
dag = test_dag
assert partial_dag.get_v_structures() == {("C", "B", "D")}
assert dag.get_v_structures() == set()
assert dag.get_v_structures(True) == {("C", "B", "D")}
assert reversed_dag.get_v_structures(True) == {("B", "D", "C")}
# Test node order doesn't change V-structure tuple order
other_dag = DAG()
other_dag.add_vertices(list("DCBA"))
other_dag.add_edges([("C", "B"), ("D", "B"), ("D", "C")])
assert other_dag.get_v_structures(True) == dag.get_v_structures(True)
def test_bif_parser():
bif_path = (Path(__file__).parent.parent / 'baynet' / 'utils' /
'bif_library' / 'earthquake.bif').resolve()
dag = dag_from_bif(bif_path)
earthquake_dag = DAG.from_modelstring(
"[Alarm|Burglary:Earthquake][Burglary][Earthquake][JohnCalls|Alarm][MaryCalls|Alarm]"
)
assert dag.nodes == earthquake_dag.nodes
assert dag.edges == earthquake_dag.edges
dag.sample(10)
with pytest.raises(ValueError):
dag = dag_from_bif("foo")
with pytest.raises(ValueError):
dag = dag_from_bif(Path("foo"))
def empty_dag() -> DAG:
return DAG.from_amat(np.zeros((4, 4)), list("ABCD"))
def partial_dag() -> DAG:
return DAG.from_modelstring("[A][B|C:D][C][D]")
def reversed_dag() -> DAG:
return DAG.from_modelstring(REVERSED_MODELSTRING)
def test_dag() -> DAG:
return DAG.from_modelstring(TEST_MODELSTRING)
def test_name_nodes():
dag = DAG(igraph.Graph(2, directed=True))
assert dag.get_node_name(0) == "A"
assert dag.get_node_index("A") == 0
