def plot_mpe_time_growing_var(bn_path):
bn = parse_network_from_file(bn_path)
variables = bn.variables
var_values = [bn.variable_values(var) for var in variables]
random_sel_values = [random.choice(val_list) for val_list in var_values]
dic = dict(zip(variables, random_sel_values))
times_list = []
for i, item in enumerate(dic):
# bn = parse_network_from_file("./sample_bayesian_networks/alarm.bif")
bn.last_max_out = None
first_n_pairs = {k: dic[k] for k in list(dic)[:i]}
ts = time.time()
mpe_ask(first_n_pairs, bn)
te = time.time()
times_list.append(te - ts)
plot_list(times_list, "MPE growing evidences", "# evidences", "time")
def test_mpe_small_bn_2(self):
small_bn = parse_network_from_file(
"./sample_bayesian_networks/earthquake.bif")
prob, assgn = mpe_ask(dict(JohnCalls="False", Alarm="True"), small_bn)
self.assertEqual(prob, 0.0006448399999999999)
self.assertEqual(assgn, [('Burglary', 'True'), ('Earthquake', 'False'),
('MaryCalls', 'True')])
def plot_mpe_chain_var_network_size():
times_list = []
for i in range(1, 500):
size = i
bn = generate_chain(size=size)
ts = time.time()
prob, assgn = mpe_ask(dict([(F"S{size}", "True")]), bn)
te = time.time()
times_list.append(te - ts)
plot_list(times_list, "MPE on chain bayesian networks",
"size of the chain", "computation time")
def test_mpe_medium_bn_2(self):
medium_bn = parse_network_from_file(
"./sample_bayesian_networks/insurance.bif")
prob, assgn = mpe_ask(dict(AntiTheft="True", RiskAversion="Cautious"),
medium_bn)
self.assertEqual(prob, 0.0007795974803453906)
self.assertEqual(assgn,
[('Age', 'Adult'), ('Mileage', 'FiftyThou'),
('SocioEcon', 'Prole'), ('OtherCar', 'True'),
('GoodStudent', 'False'), ('MakeModel', 'Economy'),
('SeniorTrain', 'False'), ('HomeBase', 'City'),
('VehicleYear', 'Older'), ('RuggedAuto', 'EggShell'),
('Antilock', 'False'), ('CarValue', 'FiveThou'),
('Airbag', 'False'), ('DrivingSkill', 'Normal'),
('Cushioning', 'Poor'), ('DrivHist', 'Zero'),
('DrivQuality', 'Normal'), ('Theft', 'False'),
('Accident', 'None'), ('OtherCarCost', 'Thousand'),
('ILiCost', 'Thousand'), ('ThisCarDam', 'None'),
('MedCost', 'Thousand'), ('ThisCarCost', 'Thousand'),
('PropCost', 'Thousand')])
def test_mpe():
bn = parse_network_from_file("./sample_bayesian_networks/earthquake.bif")
prob, assgn = mpe_ask(dict(JohnCalls="True", MaryCalls="True"), bn)
print(F"P(mpe,e): {prob}, MPE: {assgn}")
def test_mpe_medium():
bn = parse_network_from_file("./sample_bayesian_networks/insurance.bif")
prob, assgn = mpe_ask(
dict(PropCost="Thousand", RiskAversion="Adventurous"), bn)
print(F"P(mpe,e): {prob}, MPE: {assgn}")