https://github.com/mynameisvinn/basilisk

bayesian networks made easy
https://github.com/mynameisvinn/basilisk

bayesian-network causal-inference causal-models

Last synced: 14 days ago
JSON representation

bayesian networks made easy

• Host: GitHub
• URL: https://github.com/mynameisvinn/basilisk
• Owner: mynameisvinn
• Created: 2019-04-15T13:35:56.000Z (over 5 years ago)
• Default Branch: master
• Last Pushed: 2019-04-24T11:39:56.000Z (over 5 years ago)
• Last Synced: 2024-11-07T07:16:41.724Z (2 months ago)
• Topics: bayesian-network, causal-inference, causal-models
• Language: Jupyter Notebook
• Homepage:
• Size: 282 KB
• Stars: 1
• Watchers: 1
• Forks: 1
• Open Issues: 2
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

README

        
# basilisk

quickly construct bayesian networks with basilisk. define the graph and provide joint observations, and basilik will handle everything else.

## a working example

lets use [murphy's classic sprinkler example](https://www.cs.ubc.ca/~murphyk/Bayes/bayes_tutorial.pdf).

first, define nodes and their corresponding parents. in bayesian networks, a parent has a *casual* relationship with its children.

```

C = Node("cloudy")

R = Node("rain", [C])  # rain's parent is cloudy

S = Node("sprinkler", [C])

W = Node("wet", [R, S])

ls_nodes = [C, R, S, W]

```

then, instantiate a model and fit with joint observations.

```

model = BN(ls_nodes)

obs = pd.read_csv("observations.csv")  # joint observations

model.fit(obs)  # fit a model, scikit-style

model.draw_graph(node_size=1000)

```

we can inspect conditional probability tables.

```

W.cpt  # basilik automatically computes conditional probabilities

```

we can sample from any node in the graph.

```

model.execute(W)

# {'cloudy': 'True', 'sprinkler': 'False', 'rain': 'True', 'wet': 'False'}

```