added composite network to documentation

bamt/networks/base.py

@@ -655,7 +655,12 @@ def wrapper():
             c.name
             for c in self.nodes
             if type(c).__name__
-            not in ("DiscreteNode", "LogitNode", "CompositeDiscreteNode", "ConditionalLogitNode")
+            not in (
+                "DiscreteNode",
+                "LogitNode",
+                "CompositeDiscreteNode",
+                "ConditionalLogitNode",
+            )
         ]
         positive_columns = [
             c for c in cont_nodes if self.descriptor["signs"][c] == "pos"
@@ -892,7 +897,9 @@ def _encode_categorical_data(self, data):
         return data
 
     def _decode_categorical_data(self, data):
-        data = data.apply(lambda col: pd.to_numeric(col).astype(int) if col.dtype == 'object' else col)
+        data = data.apply(
+            lambda col: pd.to_numeric(col).astype(int) if col.dtype == "object" else col
+        )
         for column, encoder in self.encoders.items():
             data[column] = encoder.inverse_transform(data[column])
         return data

bamt/nodes/gaussian_node.py

@@ -97,9 +97,7 @@ def choose(self, node_info: GaussianParams, pvals: List[float]) -> float:
                 model = pickle.loads(a)
 
             if type(self).__name__ == "CompositeContinuousNode":
-                pvals = [
-                    int(item) if isinstance(item, str) else item for item in pvals
-                ]
+                pvals = [int(item) if isinstance(item, str) else item for item in pvals]
 
             cond_mean = model.predict(np.array(pvals).reshape(1, -1))[0]
             var = node_info["variance"]

bamt/nodes/logit_node.py

@@ -76,9 +76,7 @@ def choose(self, node_info: LogitParams, pvals: List[Union[float]]) -> str:
                 model = pickle.loads(a)
 
             if type(self).__name__ == "CompositeDiscreteNode":
-                pvals = [
-                    int(item) if isinstance(item, str) else item for item in pvals
-                ]
+                pvals = [int(item) if isinstance(item, str) else item for item in pvals]
             distribution = model.predict_proba(np.array(pvals).reshape(1, -1))[0]
 
             # choose

docs/source/api/networks.rst

@@ -21,6 +21,11 @@ Networks Module
    :undoc-members:
    :show-inheritance:
 
+.. automodule:: bamt.networks.composite_bn
+   :members:
+   :undoc-members:
+   :show-inheritance:
+
 .. automodule:: bamt.networks.big_brave_bn
    :members:
    :undoc-members:

docs/source/models/bayesiannetworks/composite_bn.rst

@@ -0,0 +1,63 @@
+Composite Bayesian Networks
+------------------------
+
+.. autoclass:: bamt.networks.composite_bn.CompositeBN
+   :members:
+   :no-undoc-members:
+
+Network initialization
+~~~~~~~~~~~~~~~~~~~~~~
+
+If the dataset contains both discrete and continuous variables, ``CompositeBN`` is can be used.
+To initialize a ``CompositeBN`` object, you can use the following code:
+
+.. code-block:: python
+
+    from  bamt.networks.composite_bn import CompositeBN
+
+    bn = CompositeBN()
+
+
+Data Preprocessing
+~~~~~~~~~~~~~~~~~~
+
+Before applying any structure or parametric learning, the data should be preprocessed as follows:
+
+.. code-block:: python
+
+    import bamt.Preprocessor as pp
+    import pandas as pd
+    from sklearn import preprocessing
+
+    data = pd.read_csv("path/to/data")
+    encoder = preprocessing.LabelEncoder()
+    p = pp.Preprocessor([("encoder", encoder)])
+
+    preprocessed_data, _ = p.apply(data)
+
+
+
+Structure Learning
+~~~~~~~~~~~~~~~~~~
+
+For structure learning of Composite BNs, ``bn.add_nodes()`` and ``bn.add_edges()`` methods are used.
+Data should be non-preprocessed when passed to ``bn.add_edges()``
+
+.. code-block:: python
+
+    info = p.info
+
+    bn.add_nodes(info)
+
+    bn.add_edges(data) # !!! non-preprocessed
+
+
+Parametric Learning
+~~~~~~~~~~~~~~~~~~~
+
+For parametric learning of continuous BNs, ``bn.fit_parameters()`` method is used.
+
+.. code-block:: python
+
+    bn.fit_parameters(data) # !!! non-preprocessed
+    bn.get_info()

docs/source/models/bayesiannetworks/continuous_bn.rst

@@ -1,7 +1,7 @@
 Continuous Bayesian Networks
 ----------------------------
 
-.. autoclass:: bamt.networks.ContinuousBN
+.. autoclass:: bamt.networks.continuous_bn.ContinuousBN
    :members:
    :no-undoc-members:
 
@@ -13,9 +13,9 @@ To initialize a ``ContinuousBN`` object, you can use the following code:
 
 .. code-block:: python
 
-    import bamt.Networks as Nets
+    from bamt.networks.continuous_bn import ContinuousBN
     
-    bn = Nets.ContinuousBN(use_mixture=True)
+    bn = ContinuousBN(use_mixture=True)
 
 ContinuousBN has an additional parameter ``use_mixture``. 
 It is used to determine whether to use mixuters of Gaussian distributions  to represent the conditional distribution of continuous variables.

docs/source/models/bayesiannetworks/discrete_bn.rst

@@ -13,9 +13,9 @@ To initialize a ``DiscreteBN`` object, you can use the following code:
 
 .. code-block:: python
 
-    import bamt.Networks as Nets
+    from bamt.networks.discrete_bn import DiscreteBN
 
-    bn = Nets.DiscreteBN()
+    bn = DiscreteBN()
 
 Data Preprocessing
 ~~~~~~~~~~~~~~~~~~

docs/source/models/bayesiannetworks/hybrid_bn.rst

@@ -1,7 +1,7 @@
 Hybrid Bayesian Networks
 ------------------------
 
-.. autoclass:: bamt.networks.HybridBN
+.. autoclass:: bamt.networks.hybrid_bn.HybridBN
    :members:
    :no-undoc-members:
 
@@ -13,9 +13,9 @@ To initialize a ``HybridBN`` object, you can use the following code:
 
 .. code-block:: python
 
-    import bamt.Networks as Nets
+    from  bamt.networks.hybrid_bn import HybridBN
     
-    bn = Nets.HybridBN(has_logit=True, use_mixture=True)
+    bn = HybridBN(has_logit=True, use_mixture=True)
 
 HybridBN has two additional parameters ``has_logit`` and ``use_mixture``.
 ``has_logit`` is used to determine whether to use logit nodes. Logit nodes use machine learning algorithms to represent variable.