In [4]:

    

from pgmpy.models import BayesianModel  
from pgmpy.factors.discrete import TabularCPD


    

In [5]:

    

model = BayesianModel()


    

In [6]:

    

model.add_nodes_from(["accident", "rain", "traffic_jam"])
model.add_edge("rain", "traffic_jam")
model.add_edge("accident", "traffic_jam")


    

In [7]:

    

model.nodes()


    

    
Out[7]:





NodeView(('accident', 'rain', 'traffic_jam'))

In [8]:

    

model.edges()


    

    
Out[8]:





OutEdgeView([('accident', 'traffic_jam'), ('rain', 'traffic_jam')])

In [9]:

    

cpd_rain = TabularCPD("rain", variable_card=2, values=[[0.4, 0.6]])
cpd_accident = TabularCPD("accident", variable_card=2, values=[[0.1, 0.9]])
cpd_traffic_jam = TabularCPD(
    "traffic_jam",
    variable_card=2,
    values=[
        [0.9, 0.6, 0.7, 0.1],
        [0.1, 0.4, 0.3, 0.9]
    ],
    evidence=["rain", "accident"],
    evidence_card=[2, 2])

model.add_cpds(cpd_rain, cpd_accident, cpd_traffic_jam)
model.get_cpds()


    

    
Out[9]:





[<TabularCPD representing P(rain:2) at 0x7f5c9a888e50>,
 <TabularCPD representing P(accident:2) at 0x7f5c9a888e10>,
 <TabularCPD representing P(traffic_jam:2 | rain:2, accident:2) at 0x7f5c9a888ed0>]

In [10]:

    

model.check_model()


    

    
Out[10]:





True

In [11]:

    

model.get_independencies()


    

    
Out[11]:





(accident _|_ rain)
(rain _|_ accident)

In [12]:

    

model.is_active_trail("accident", "rain")


    

    
Out[12]:





False

In [15]:

    

model.is_active_trail("accident", "rain", observed="traffic_jam")


    

    
Out[15]:





True

In [44]:

    

import networkx as nx
import matplotlib.pyplot as plt
nx.draw(
    model,
    with_labels=True,
    node_size=2000, 
    node_color="skyblue", 
    node_shape="o", 
    alpha=0.7,
    linewidths=5
)
plt.show()


    

In [20]:

    

from pgmpy.inference import VariableElimination
inference = VariableElimination(model)

# Computing the probability of bronc given smoke.
q = inference.query(variables=["traffic_jam"], evidence={"rain": 0, "accident": 1})
print(q)


    

    




Finding Elimination Order: : : 0it [00:00, ?it/s]
0it [00:00, ?it/s]





    




+----------------+--------------------+
| traffic_jam    |   phi(traffic_jam) |
+================+====================+
| traffic_jam(0) |             0.6000 |
+----------------+--------------------+
| traffic_jam(1) |             0.4000 |
+----------------+--------------------+






    

In [ ]: