def test_construct_bayes_net() -> None:
uniform = UniformInt(0, 1)
graph = set(uniform.get_connected_graph())
vertex_ids = [vertex.get_id() for vertex in graph]
assert len(vertex_ids) == 3
assert uniform.get_id() in vertex_ids
net = BayesNet(graph)
latent_vertex_ids = [
vertex.get_id() for vertex in net.get_latent_vertices()
]
assert len(latent_vertex_ids) == 1
assert uniform.get_id() in latent_vertex_ids
def model(self) -> Model:
start_year, end_year = (self._data.index.min(), self._data.index.max())
with Model() as m:
m.switchpoint = UniformInt(start_year, end_year + 1)
m.early_rate = Exponential(1.0)
m.late_rate = Exponential(1.0)
m.years = np.array(self._data.index)
m.rates = If(m.switchpoint > m.years, m.early_rate, m.late_rate)
m.disasters = Poisson(m.rates)
return m
python_id = gaussian.get_id()
assert all(value in python_id for value in java_id)
def test_ids_are_reset() -> None:
gaussian = Gaussian(0., 1.)
set_deterministic_state()
gaussian2 = Gaussian(0., 1.)
assert gaussian.get_id() == gaussian2.get_id()
@pytest.mark.parametrize("vertex, expected_type",
[(Gaussian(0., 1.), np.floating),
(UniformInt(0, 10), np.integer),
(Bernoulli(0.5), np.bool_)])
@pytest.mark.parametrize(
"value, assert_vertex_value_equals",
[(np.array([[4]]), assert_vertex_value_equals_ndarray),
(np.array([[5.]]), assert_vertex_value_equals_ndarray),
(np.array([[True]]), assert_vertex_value_equals_ndarray),
(np.array([[1, 2], [3, 4]]), assert_vertex_value_equals_ndarray),
(pd.Series(data=[4]), assert_vertex_value_equals_pandas),
(pd.Series(data=[5.]), assert_vertex_value_equals_pandas),
(pd.Series(data=[True]), assert_vertex_value_equals_pandas),
(pd.Series(data=[1, 2, 3]), assert_vertex_value_equals_pandas),
(pd.Series(data=[1., 2., 3.]), assert_vertex_value_equals_pandas),
(pd.Series(data=[True, False, False]), assert_vertex_value_equals_pandas),
(pd.DataFrame(data=[[4]]), assert_vertex_value_equals_pandas),
(pd.DataFrame(data=[[5.]]), assert_vertex_value_equals_pandas),