Add a method to export decision paths

treefarms/model/treefarms.py

@@ -31,7 +31,7 @@ def __init__(self, configuration={}):
         self.model_set = None
         self.dataset = None
 
-    # TODO: implement this 
+    # TODO: implement this
     def load(self, path):
         """
         Parameters
@@ -87,7 +87,7 @@ def __train__(self, X, y):
         self.model_set = ModelSetContainer(result)
 
         print(f"training completed. Number of trees in the Rashomon set: {self.model_set.get_tree_count()}")
-    
+
 
     def fit(self, X, y):
         """
@@ -104,7 +104,7 @@ def fit(self, X, y):
         self.__train__(X, y)
         return self
 
-    # TODO: implement this 
+    # TODO: implement this
     def predict(self, X):
         """
         Parameters
@@ -133,7 +133,7 @@ def __getitem__(self, idx):
         if self.model_set is None:
             raise Exception("Error: Model not yet trained")
         return self.model_set.__getitem__(idx)
-        
+
     def get_tree_count(self):
         """Returns the number of trees in the Rashomon set
 
@@ -145,30 +145,47 @@ def get_tree_count(self):
         if self.model_set is None:
             raise Exception("Error: Model not yet trained")
         return self.model_set.get_tree_count()
-
-    def visualize(self, feature_names=None, feature_description=None, *, width=500, height=650):
-        """Generates a visualization of the Rashomon set using `timbertrek`
+
+    def get_decision_paths(self, feature_names=None, feature_description=None):
+        """Create a hierarchical dictionary describing the decision paths in the
+        Rashomon set using `timbertrek`.
         Parameters
         ---
         feature_names : matrix-like, shape = [m_features + 1]
             a matrix where each row is a sample to be predicted and each column is a feature to be used for prediction
         """
         if self.model_set is None:
             raise Exception("Error: Model not yet trained")
+
+        # Convert the trie structure to decision paths
         trie = self.model_set.to_trie()
         df = self.dataset
         if feature_names is None:
             feature_names = df.columns
-        
+
         decision_paths = timbertrek.transform_trie_to_rules(
             trie,
             df,
             feature_names=feature_names,
             feature_description=feature_description,
         )
-
-        # return decision_paths
-
+
+        return decision_paths
+
+    def visualize(self, feature_names=None, feature_description=None, *, width=500, height=650):
+        """Generates a visualization of the Rashomon set using `timbertrek`
+        Parameters
+        ---
+        feature_names : matrix-like, shape = [m_features + 1]
+            a matrix where each row is a sample to be predicted and each column is a feature to be used for prediction
+        """
+        # Get the decision paths
+        decision_paths = self.get_decision_paths(
+            feature_names=feature_names,
+            feature_description=feature_description
+        )
+
+        # Show in the in-notebook visualization
         timbertrek.visualize(decision_paths, width=width, height=height)
 
     def __translate__(self, leaves):