def test__generate_mutual_information_graph(self):
samples = [
[0, 0, 0],
[0, 0, 1],
[0, 1, 1],
[1, 1, 1],
]
expected_results = [
(0, 1, {
'weight': -0.21576155433883565
}),
(0, 2, {
'weight': -0.084949518397698542
}),
(1, 2, {
'weight': -0.21576155433883565
}),
]
distribution = mimic.Distribution(samples)
self.assertEqual(
expected_results,
distribution.complete_graph.edges(data=True),
)
def test__generate_spanning_graph(self):
samples = [
[0, 0, 0],
[0, 0, 1],
[0, 1, 1],
[1, 1, 1],
]
expected_results = [
(0, 1),
(1, 2),
]
distribution = mimic.Distribution(samples)
self.assertEqual(
expected_results,
distribution.spanning_graph.edges(),
)
def test_generate_samples(self):
expected_results = np.array([[1, 0, 1, 1]])
graph = nx.DiGraph()
graph.add_node(0, probabilities={0: 0, 1: 1})
graph.add_node(1, probabilities={0: {0: 0, 1: 1}, 1: {0: 1, 1: 0}})
graph.add_node(2, probabilities={0: {0: 0, 1: 1}, 1: {0: 1, 1: 0}})
graph.add_node(3, probabilities={0: {0: 0, 1: 1}, 1: {0: 1, 1: 0}})
graph.add_edges_from([
(0, 1),
(1, 2),
(1, 3),
])
distribution = mimic.Distribution([[0, 0], [0, 0]])
distribution.bayes_net = graph
self.assertTrue(
np.equal(
expected_results,
distribution.generate_samples(1),
).all())