fix for static plot showing the weights correctly in the edges

bnlearn/__init__.py

@@ -38,7 +38,7 @@
 
 __author__ = 'Erdogan Tasksen'
 __email__ = 'erdogant@gmail.com'
-__version__ = '0.8.6'
+__version__ = '0.8.7'
 
 import pgmpy
 # Check version pgmpy

bnlearn/bnlearn.py

@@ -964,7 +964,7 @@ def plot(model,
          node_properties=None,
          edge_properties=None,
          params_interactive={'minmax_distance': [100, 250], 'figsize': (1500, 800), 'notebook': False, 'font_color': 'node_color', 'bgcolor': '#ffffff', 'show_slider': True, 'filepath': None},
-         params_static={'minscale': 1, 'maxscale': 10, 'figsize': (15, 10), 'width': None, 'height': None, 'font_size': 14, 'font_family': 'sans-serif', 'alpha': 0.8, 'node_shape': 'o', 'layout': 'spring_layout', 'font_color': '#000000', 'facecolor': 'white', 'edge_alpha': 0.8, 'arrowstyle': '-|>', 'arrowsize': 30, 'visible': True},
+         params_static={'minscale': 1, 'maxscale': 10, 'figsize': (15, 10), 'width': None, 'height': None, 'font_size': 14, 'font_family': 'sans-serif', 'alpha': 0.8, 'node_shape': 'o', 'layout': 'spectral_layout', 'font_color': '#000000', 'facecolor': 'white', 'edge_alpha': 0.8, 'arrowstyle': '-|>', 'arrowsize': 30, 'visible': True},
          verbose=3):
     """Plot the learned stucture.
 
@@ -1050,7 +1050,7 @@ def plot(model,
     # Plot properties
     defaults = {'minmax_distance': [100, 250], 'figsize': (1500, 800), 'notebook': False, 'font_color': 'node_color', 'bgcolor': '#ffffff', 'directed': True, 'show_slider': True, 'filepath': None}
     params_interactive = {**defaults, **params_interactive}
-    defaults = {'minscale': 1, 'maxscale': 10, 'figsize': (15, 10), 'height': None, 'width': None, 'font_size': 14, 'font_family': 'sans-serif', 'alpha': 0.8, 'layout': 'spring_layout', 'font_color': 'k', 'facecolor': '#ffffff', 'node_shape': 'o', 'edge_alpha': 0.8, 'arrowstyle': '-|>', 'arrowsize': 30, 'visible': True}
+    defaults = {'minscale': 1, 'maxscale': 10, 'figsize': (15, 10), 'height': None, 'width': None, 'font_size': 14, 'font_family': 'sans-serif', 'alpha': 0.8, 'layout': 'spectral_layout', 'font_color': 'k', 'facecolor': '#ffffff', 'node_shape': 'o', 'edge_alpha': 0.8, 'arrowstyle': '-|>', 'arrowsize': 30, 'visible': True}
     params_static = {**defaults, **params_static}
 
     # DEPRECATED IN LATER VERSION
@@ -1075,6 +1075,15 @@ def plot(model,
         if node_properties[key]['node_size'] is None:
             node_properties[key]['node_size']=node_size_default
 
+    # Add edges with weights based on independence test results
+    for edge, properties in edge_properties.items():
+        strength = properties.get("weight", 0)
+        G.add_edge(*edge, weight = strength)
+
+    # Update the dataframe with the weights
+    for (source, target), value in edge_properties.items():
+        model['adjmat'].loc[source, target] = value['weight']
+
     # Extract model if in dict
     if 'dict' in str(type(model)):
         bnmodel = model.get('model', None)
@@ -1110,7 +1119,7 @@ def plot(model,
         if ('bayes' in str(type(bnmodel)).lower()) or ('pgmpy' in str(type(bnmodel)).lower()):
             if verbose>=3: print('[bnlearn] >Plot based on Bayesian model')
             # positions for all nodes
-            G = nx.DiGraph(model['adjmat'])
+            # G = nx.DiGraph(model['adjmat'])
             pos = bnlearn.network.graphlayout(G, pos=pos, scale=scale, layout=params_static['layout'], verbose=verbose)
         elif 'networkx' in str(type(bnmodel)):
             if verbose>=3: print('[bnlearn] >Plot based on networkx model')

bnlearn/examples.py

@@ -1,3 +1,31 @@
+# %% issue #93
+
+import bnlearn as bn
+
+# Load example mixed dataset
+df = bn.import_example(data='titanic')
+
+# Convert to onehot
+dfhot, dfnum = bn.df2onehot(df)
+
+# Structure learning
+# model = bn.structure_learning.fit(dfnum, methodtype='cl', black_list=['Embarked','Parch','Name'], root_node='Survived', bw_list_method='nodes')
+model = bn.structure_learning.fit(dfnum)
+# Plot
+G = bn.plot(model, interactive=False)
+
+# Compute edge strength with the chi_square test statistic
+model = bn.independence_test(model, dfnum, test='chi_square', prune=True)
+
+# Plot
+bn.plot(model, interactive=False, pos=G['pos'], params_static={'layout': 'spectral_layout'})
+# 'spring_layout', 'planar_layout', 'shell_layout', 'spectral_layout', 'pydot_layout', 'graphviz_layout', 'circular_layout', 'spring_layout', 'random_layout', 'bipartite_layout', 'multipartite_layout',
+# bn.plot(model, interactive=True, pos=G['pos'])
+
+# Parameter learning
+model = bn.parameter_learning.fit(model, dfnum)
+
+
 # %% compute causalities
 import bnlearn as bn
 # Load asia DAG