class TestDynamicBayesianNetworkCreation(unittest.TestCase):
def setUp(self):
self.network = DynamicBayesianNetwork()
def test_add_single_node(self):
self.network.add_node('a')
self.assertListEqual(self.network.nodes(), ['a'])
def test_add_multiple_nodes(self):
self.network.add_nodes_from(['a', 'b', 'c'])
self.assertListEqual(sorted(self.network.nodes()), ['a', 'b', 'c'])
def test_add_single_edge_with_timeslice(self):
self.network.add_edge(('a', 0), ('b', 0))
self.assertListEqual(sorted(self.network.edges()), [(('a', 0), ('b', 0)), (('a', 1), ('b', 1))])
self.assertListEqual(sorted(self.network.nodes()), ['a', 'b'])
def test_add_edge_with_different_number_timeslice(self):
self.network.add_edge(('a', 2), ('b', 2))
self.assertListEqual(sorted(self.network.edges()), [(('a', 0), ('b', 0)), (('a', 1), ('b', 1))])
def test_add_edge_going_backward(self):
self.assertRaises(NotImplementedError, self.network.add_edge, ('a', 1), ('b', 0))
def test_add_edge_with_farther_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 2), ('b', 4))
def test_add_edge_with_self_loop(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 0), ('a', 0))
def test_add_edge_with_varying_length(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 1, 1), ('b', 2))
self.assertRaises(ValueError, self.network.add_edge, ('b', 2), ('a', 2, 3))
def test_add_edge_with_closed_path(self):
self.assertRaises(ValueError, self.network.add_edges_from,
[(('a', 0), ('b', 0)), (('b', 0), ('c', 0)), (('c', 0), ('a', 0))])
def test_add_single_edge_without_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, 'a', 'b')
def test_add_single_edge_with_incorrect_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 'b'), ('b', 'c'))
def test_add_multiple_edges(self):
self.network.add_edges_from([(('a', 0), ('b', 0)), (('a', 0), ('a', 1)), (('b', 0), ('b', 1))])
self.assertListEqual(sorted(self.network.edges()),
[(('a', 0), ('a', 1)), (('a', 0), ('b', 0)), (('a', 1), ('b', 1)), (('b', 0), ('b', 1))])
def tearDown(self):
del self.network
class TestDynamicBayesianNetworkCreation(unittest.TestCase):
def setUp(self):
self.network = DynamicBayesianNetwork()
def test_add_single_node(self):
self.network.add_node('a')
self.assertListEqual(self.network.nodes(), ['a'])
def test_add_multiple_nodes(self):
self.network.add_nodes_from(['a', 'b', 'c'])
self.assertListEqual(sorted(self.network.nodes()), ['a', 'b', 'c'])
def test_add_single_edge_with_timeslice(self):
self.network.add_edge(('a', 0), ('b', 0))
self.assertListEqual(sorted(self.network.edges()), [(('a', 0), ('b', 0)), (('a', 1), ('b', 1))])
self.assertListEqual(sorted(self.network.nodes()), ['a', 'b'])
def test_add_edge_with_different_number_timeslice(self):
self.network.add_edge(('a', 2), ('b', 2))
self.assertListEqual(sorted(self.network.edges()), [(('a', 0), ('b', 0)), (('a', 1), ('b', 1))])
def test_add_edge_going_backward(self):
self.assertRaises(NotImplementedError, self.network.add_edge, ('a', 1), ('b', 0))
def test_add_edge_with_farther_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 2), ('b', 4))
def test_add_edge_with_self_loop(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 0), ('a', 0))
def test_add_edge_with_varying_length(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 1, 1), ('b', 2))
self.assertRaises(ValueError, self.network.add_edge, ('b', 2), ('a', 2, 3))
def test_add_edge_with_closed_path(self):
self.assertRaises(ValueError, self.network.add_edges_from,
[(('a', 0), ('b', 0)), (('b', 0), ('c', 0)), (('c', 0), ('a', 0))])
def test_add_single_edge_without_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, 'a', 'b')
def test_add_single_edge_with_incorrect_timeslice(self):
self.assertRaises(ValueError, self.network.add_edge, ('a', 'b'), ('b', 'c'))
def test_add_multiple_edges(self):
self.network.add_edges_from([(('a', 0), ('b', 0)), (('a', 0), ('a', 1)), (('b', 0), ('b', 1))])
self.assertListEqual(sorted(self.network.edges()),
[(('a', 0), ('a', 1)), (('a', 0), ('b', 0)), (('a', 1), ('b', 1)), (('b', 0), ('b', 1))])
def tearDown(self):
del self.network
import numpy as np
import pandas as pd
import sys
from pgmpy.models import DynamicBayesianNetwork
from pgmpy.estimators import BayesianEstimator
from pgmpy.inference import VariableElimination
import matplotlib.pyplot as plt
model = DynamicBayesianNetwork()
list_edges = [(('DPQ', 0), ('DI', 0)), (('C', 0), ('DI', 0))]
for i in range(3):
list_edges += [(('DI', i), ('DFT', i)),
(('TQ', i), ('DFT', i)),
(('DFT', i), ('RD', i)),
(('RD', i), ('DFO', i)),
(('OU', i), ('DFO', i))]
if (i == 2):
break
list_edges += [(('RD', i), ('DI', i + 1)),
(('TQ', i), ('TQ', i + 1)),
(('OU', i), ('OU', i + 1))]
model.add_edges_from(list_edges)
print(model.edges())
print(model.nodes())
class TestDynamicBayesianNetworkMethods(unittest.TestCase):
def setUp(self):
self.network = DynamicBayesianNetwork()
self.grade_cpd = TabularCPD(('G', 0), 3, values=[[0.3, 0.05, 0.8, 0.5],
[0.4, 0.25, 0.1, 0.3],
[0.3, 0.7, 0.1, 0.2]],
evidence=[('D', 0), ('I', 0)], evidence_card=[2, 2])
self.d_i_cpd = TabularCPD(('D', 1), 2, values=[[0.6, 0.3], [0.4, 0.7]],
evidence=[('D', 0)], evidence_card=2)
self.diff_cpd = TabularCPD(('D', 0), 2, values=[[0.6, 0.4]])
self.intel_cpd = TabularCPD(('I', 0), 2, values=[[0.7, 0.3]])
self.i_i_cpd = TabularCPD(('I', 1), 2, values=[[0.5, 0.4], [0.5, 0.6]],
evidence=[('I', 0)], evidence_card=[2])
self.grade_1_cpd = TabularCPD(('G', 1), 3, values=[[0.3, 0.05, 0.8, 0.5],
[0.4, 0.25, 0.1, 0.3],
[0.3, 0.7, 0.1, 0.2]],
evidence=[('D', 1), ('I', 1)], evidence_card=[2, 2])
def test_get_intra_and_inter_edges(self):
self.network.add_edges_from([(('a', 0), ('b', 0)), (('a', 0), ('a', 1)), (('b', 0), ('b', 1))])
self.assertListEqual(sorted(self.network.get_intra_edges()), [(('a', 0), ('b', 0))])
self.assertListEqual(sorted(self.network.get_intra_edges(1)), [(('a', 1), ('b', 1))])
self.assertRaises(ValueError, self.network.get_intra_edges, -1)
self.assertRaises(ValueError, self.network.get_intra_edges, '-')
self.assertListEqual(sorted(self.network.get_inter_edges()), [(('a', 0), ('a', 1)), (('b', 0), ('b', 1))])
def test_get_interface_nodes(self):
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
self.assertListEqual(sorted(self.network.get_interface_nodes()), [('D', 0), ('I', 0)])
self.assertRaises(ValueError, self.network.get_interface_nodes, -1)
self.assertRaises(ValueError, self.network.get_interface_nodes, '-')
def test_get_slice_nodes(self):
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
self.assertListEqual(sorted(self.network.get_slice_nodes()), [('D', 0), ('G', 0), ('I', 0)])
self.assertListEqual(sorted(self.network.get_slice_nodes(1)), [('D', 1), ('G', 1), ('I', 1)])
self.assertRaises(ValueError, self.network.get_slice_nodes, -1)
self.assertRaises(ValueError, self.network.get_slice_nodes, '-')
def test_add_single_cpds(self):
self.network.add_edges_from([(('D', 0), ('G', 0)), (('I', 0), ('G', 0))])
self.network.add_cpds(self.grade_cpd)
self.assertListEqual(self.network.get_cpds(), [self.grade_cpd])
def test_get_cpds(self):
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
self.network.add_cpds(self.grade_cpd, self.d_i_cpd, self.diff_cpd, self.intel_cpd, self.i_i_cpd)
self.network.initialize_initial_state()
self.assertEqual(set(self.network.get_cpds()), set([self.diff_cpd, self.intel_cpd, self.grade_cpd]))
self.assertEqual(self.network.get_cpds(time_slice=1)[0].variable, ('G', 1))
def test_add_multiple_cpds(self):
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
self.network.add_cpds(self.grade_cpd, self.d_i_cpd, self.diff_cpd, self.intel_cpd, self.i_i_cpd)
self.assertEqual(self.network.get_cpds(('G', 0)).variable, ('G', 0))
self.assertEqual(self.network.get_cpds(('D', 1)).variable, ('D', 1))
self.assertEqual(self.network.get_cpds(('D', 0)).variable, ('D', 0))
self.assertEqual(self.network.get_cpds(('I', 0)).variable, ('I', 0))
self.assertEqual(self.network.get_cpds(('I', 1)).variable, ('I', 1))
def test_initialize_initial_state(self):
self.network.add_nodes_from(['D', 'G', 'I', 'S', 'L'])
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
self.network.add_cpds(self.grade_cpd, self.d_i_cpd, self.diff_cpd, self.intel_cpd, self.i_i_cpd)
self.network.initialize_initial_state()
self.assertEqual(len(self.network.cpds), 6)
self.assertEqual(self.network.get_cpds(('G', 1)).variable, ('G', 1))
def test_moralize(self):
self.network.add_edges_from(([(('D', 0), ('G', 0)), (('I', 0), ('G', 0))]))
moral_graph = self.network.moralize()
self.assertListEqual(hf.recursive_sorted(moral_graph.edges()),
[[('D', 0), ('G', 0)], [('D', 0), ('I', 0)],
[('D', 1), ('G', 1)], [('D', 1), ('I', 1)],
[('G', 0), ('I', 0)], [('G', 1), ('I', 1)]])
def test_copy(self):
self.network.add_edges_from(
[(('D', 0), ('G', 0)), (('I', 0), ('G', 0)), (('D', 0), ('D', 1)), (('I', 0), ('I', 1))])
cpd = TabularCPD(('G', 0), 3, values=[[0.3, 0.05, 0.8, 0.5],
[0.4, 0.25, 0.1, 0.3],
[0.3, 0.7, 0.1, 0.2]],
evidence=[('D', 0), ('I', 0)], evidence_card=[2, 2])
self.network.add_cpds(cpd)
copy = self.network.copy()
self.assertIsInstance(copy, DynamicBayesianNetwork)
self.assertListEqual(sorted(self.network.nodes()), sorted(copy.nodes()))
self.assertListEqual(sorted(self.network.edges()), sorted(copy.edges()))
self.assertListEqual(self.network.get_cpds(), copy.get_cpds())
self.assertListEqual(sorted(self.network.get_intra_edges()), sorted(copy.get_intra_edges()))
self.assertListEqual(sorted(self.network.get_inter_edges()), sorted(copy.get_inter_edges()))
self.assertListEqual(sorted(self.network.get_slice_nodes()), sorted(copy.get_slice_nodes()))
copy.cpds[0].values = np.array([[0.4, 0.05, 0.3, 0.5], [0.3, 0.25, 0.5, 0.3], [0.3, 0.7, 0.2, 0.2]])
self.assertNotEqual(self.network.get_cpds(), copy.get_cpds())
self.network.add_cpds(self.i_i_cpd, self.d_i_cpd)
copy.add_cpds(self.diff_cpd, self.intel_cpd)
self.network.add_node('A')
copy.add_node('Z')
self.network.add_edge(('A', 0), ('D', 0))
copy.add_edge(('Z', 0), ('D', 0))
self.assertNotEqual(sorted(self.network.nodes()), sorted(copy.nodes()))
self.assertNotEqual(sorted(self.network.edges()), sorted(copy.edges()))
self.assertNotEqual(self.network.get_cpds(), copy.get_cpds())
self.assertNotEqual(sorted(self.network.get_intra_edges()), sorted(copy.get_intra_edges()))
self.assertListEqual(sorted(self.network.get_inter_edges()), sorted(copy.get_inter_edges()))
self.assertNotEqual(sorted(self.network.get_slice_nodes()), sorted(copy.get_slice_nodes()))
self.network.add_edge(('A', 0), ('D', 1))
copy.add_edge(('Z', 0), ('D', 1))
self.assertNotEqual(sorted(self.network.get_inter_edges()), sorted(copy.get_inter_edges()))
def tearDown(self):
del self.network
def estimate_dynamic(self, start=None, tabu_length=0, max_indegree=None):
"""
Performs local hill climb search to estimates the `BayesianModel` structure
that has optimal score, according to the scoring method supplied in the constructor.
Starts at model `start` and proceeds by step-by-step network modifications
until a local maximum is reached. Only estimates network structure, no parametrization.
Parameters
----------
start: DynamicBayesianNetwork instance
The starting point for the local search. By default a completely disconnected network is used.
tabu_length: int
If provided, the last `tabu_length` graph modifications cannot be reversed
during the search procedure. This serves to enforce a wider exploration
of the search space. Default value: 100.
max_indegree: int or None
If provided and unequal None, the procedure only searches among models
where all nodes have at most `max_indegree` parents. Defaults to None.
Returns
-------
model: `DynamicBayesianNetwork` instance
A `DynamicBayesianModel` at a (local) score maximum.
Examples
--------
>>> import pandas as pd
>>> import numpy as np
>>> from pgmpy.estimators import HillClimbSearch, BicScore
>>> from pgmpy.models import DynamicBayesianNetwork as DBN
>>> # create data sample with 9 random variables:
... data = pd.DataFrame(np.random.randint(0, 5, size=(5000, 9)), c olumns=list('ABCDEFGHI'))
>>> # add 10th dependent variable
... data['J'] = data['A'] * data['B']
>>> labels = np.array(data.columns)
>>> transitionModel = DBN()
>>> transitionModel.add_nodes_from(labels)
>>> est = HillClimbSearch(data, labels, scoring_method=BicScore(data, labels))
>>> best_model = est.estimate_dynamic(start=transitionModel)
>>> sorted(best_model.nodes())
['A', 'B', 'C', 'D', 'E', 'F', 'G', 'H', 'I', 'J']
>>> best_model.edges()
[(('A', 1), ('J', 1)), (('A', 1), ('B', 1)), (('J', 1), ('B', 1))]
"""
epsilon = 1e-8
if self.nodes is None:
nodes = self.state_names.keys()
else:
nodes = self.nodes
if start is None:
start = DB()
start.add_nodes_from(nodes)
elif not isinstance(start, DB) or not set(start.nodes()) == set(nodes):
raise ValueError(
"'start' should be a DynamicBayesianModel with the same variables as the data set, or 'None'."
)
tabu_list = []
current_model = start
while True:
best_score_delta = 0
best_operation = None
for operation, score_delta in self._legal_operations_dynamic(
current_model, tabu_list, max_indegree):
if score_delta > best_score_delta:
best_operation = operation
best_score_delta = score_delta
if best_operation is None or best_score_delta < epsilon:
break
elif best_operation[0] == '+':
current_model.add_edges_from([(best_operation[1][0],
best_operation[1][1])])
tabu_list = ([('-', best_operation[1])] +
tabu_list)[:tabu_length]
elif best_operation[0] == '-':
current_model.remove_edge((best_operation[1][0]),
(best_operation[1][1]))
tabu_list = ([('+', best_operation[1])] +
tabu_list)[:tabu_length]
elif best_operation[0] == 'flip':
((X, A), (Y, B)) = best_operation[1]
current_model.remove_edge((X, A), (Y, B))
current_model.add_edge((Y, A), (X, B))
tabu_list = ([best_operation] + tabu_list)[:tabu_length]
return current_model
