update FCI test, refactor minor parts, update doc

Author: kunwuz

causallearn/search/ConstraintBased/FCI.py

@@ -553,7 +553,7 @@ def ruleR4B(graph, maxPathLength, data, independence_test_method, alpha, sep_set
 
 
 def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_path_length=-1,
-        verbose=False, knowledge=None):
+        verbose=False, background_knowledge=None):
     '''
     Causal Discovery with Fast Causal Inference
 
@@ -565,7 +565,7 @@ def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_p
     depth: The depth for the fast adjacency search, or -1 if unlimited
     max_path_length: the maximum length of any discriminating path, or -1 if unlimited.
     verbose: True is verbose output should be printed or logged
-    knowledge: background knowledge
+    background_knowledge: background knowledge
 
     Returns
     -------
@@ -576,8 +576,8 @@ def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_p
     ## ------- check parameters ------------
     if (depth is None) or type(depth) != int:
         raise TypeError("'depth' must be 'int' type!")
-    if (knowledge is not None) and type(knowledge) != BackgroundKnowledge:
-        raise TypeError("'knowledge' must be 'BackgroundKnowledge' type!")
+    if (background_knowledge is not None) and type(background_knowledge) != BackgroundKnowledge:
+        raise TypeError("'background_knowledge' must be 'BackgroundKnowledge' type!")
     if (max_path_length is not None) and type(max_path_length) != int:
         raise TypeError("'max_path_length' must be 'int' type!")
     ## ------- end check parameters ------------
@@ -588,7 +588,7 @@ def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_p
         node.add_attribute("id", i)
         nodes.append(node)
 
-    graph, sep_sets = fas(dataset, nodes, independence_test_method=independence_test_method, alpha=alpha, knowledge=knowledge, depth=depth, verbose=verbose)
+    graph, sep_sets = fas(dataset, nodes, independence_test_method=independence_test_method, alpha=alpha, knowledge=background_knowledge, depth=depth, verbose=verbose)
 
     # reorient all edges with CIRCLE Endpoint
     ori_edges = graph.get_graph_edges()
@@ -598,9 +598,9 @@ def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_p
         ori_edge.set_endpoint2(Endpoint.CIRCLE)
         graph.add_edge(ori_edge)
 
-    sp = SepsetsPossibleDsep(dataset, graph, independence_test_method, alpha, knowledge, depth, max_path_length, verbose)
+    sp = SepsetsPossibleDsep(dataset, graph, independence_test_method, alpha, background_knowledge, depth, max_path_length, verbose)
 
-    rule0(graph, nodes, sep_sets, knowledge, verbose)
+    rule0(graph, nodes, sep_sets, background_knowledge, verbose)
 
     waiting_to_deleted_edges = []
 
@@ -626,21 +626,21 @@ def fci(dataset, independence_test_method = fisherz, alpha=0.05, depth=-1, max_p
             print(message)
 
     reorientAllWith(graph, Endpoint.CIRCLE)
-    rule0(graph, nodes, sep_sets, knowledge, verbose)
+    rule0(graph, nodes, sep_sets, background_knowledge, verbose)
 
     changeFlag = True
     firstTime = True
 
     while changeFlag:
         changeFlag = False
-        changeFlag = rulesR1R2cycle(graph, knowledge, changeFlag, verbose)
-        changeFlag = ruleR3(graph, sep_sets, knowledge, changeFlag, verbose)
-
-        if changeFlag or (firstTime and knowledge is not None and
-                          len(knowledge.forbidden_rules_specs) > 0 and
-                          len(knowledge.required_rules_specs) > 0 and
-                          len(knowledge.tier_map.keys()) > 0):
-            changeFlag = ruleR4B(graph, max_path_length, dataset, independence_test_method, alpha, sep_sets, changeFlag, knowledge, verbose)
+        changeFlag = rulesR1R2cycle(graph, background_knowledge, changeFlag, verbose)
+        changeFlag = ruleR3(graph, sep_sets, background_knowledge, changeFlag, verbose)
+
+        if changeFlag or (firstTime and background_knowledge is not None and
+                          len(background_knowledge.forbidden_rules_specs) > 0 and
+                          len(background_knowledge.required_rules_specs) > 0 and
+                          len(background_knowledge.tier_map.keys()) > 0):
+            changeFlag = ruleR4B(graph, max_path_length, dataset, independence_test_method, alpha, sep_sets, changeFlag, background_knowledge, verbose)
 
             firstTime = False
 

causallearn/search/FCMBased/lingam/base.py

@@ -124,7 +124,7 @@ def _estimate_adjacency_matrix(self, X, prior_knowledge=None):
             Training data, where n_samples is the number of samples
             and n_features is the number of features.
         prior_knowledge : array-like, shape (n_variables, n_variables), optional (default=None)
-            Prior knowledge matrix.
+            Prior background_knowledge matrix.
 
         Returns
         -------
@@ -140,7 +140,7 @@ def _estimate_adjacency_matrix(self, X, prior_knowledge=None):
             target = self._causal_order[i]
             predictors = self._causal_order[:i]
 
-            # Exclude variables specified in no_path with prior knowledge
+            # Exclude variables specified in no_path with prior background_knowledge
             if prior_knowledge is not None:
                 predictors = [p for p in predictors if pk[target, p] != 0]
 

causallearn/search/FCMBased/lingam/bottom_up_parce_lingam.py

@@ -39,13 +39,13 @@ def __init__(self, random_state=None, alpha=0.1, regressor=None, prior_knowledge
             Regressor to compute residuals.
             This regressor object must have ``fit``method and ``predict`` function like scikit-learn's model.
         prior_knowledge : array-like, shape (n_features, n_features), optional (default=None)
-            Prior knowledge used for causal discovery, where ``n_features`` is the number of features.
+            Prior background_knowledge used for causal discovery, where ``n_features`` is the number of features.
 
-            The elements of prior knowledge matrix are defined as follows [1]_:
+            The elements of prior background_knowledge matrix are defined as follows [1]_:
 
             * ``0`` : :math:`x_i` does not have a directed path to :math:`x_j`
             * ``1`` : :math:`x_i` has a directed path to :math:`x_j`
-            * ``-1`` : No prior knowledge is available to know if either of the two cases above (0 or 1) is true.
+            * ``-1`` : No prior background_knowledge is available to know if either of the two cases above (0 or 1) is true.
         """
         # Check parameters
         if regressor is not None:
@@ -66,7 +66,7 @@ def __init__(self, random_state=None, alpha=0.1, regressor=None, prior_knowledge
             self._Aknw = check_array(self._Aknw)
             self._Aknw = np.where(self._Aknw < 0, np.nan, self._Aknw)
 
-            # Extract all partial orders in prior knowledge matrix
+            # Extract all partial orders in prior background_knowledge matrix
             self._partial_orders = self._extract_partial_orders(self._Aknw)
 
     def fit(self, X):
@@ -90,11 +90,11 @@ def fit(self, X):
         X = check_array(X)
         n_features = X.shape[1]
 
-        # Check prior knowledge
+        # Check prior background_knowledge
         if self._Aknw is not None:
             if (n_features, n_features) != self._Aknw.shape:
                 raise ValueError(
-                    'The shape of prior knowledge must be (n_features, n_features)')
+                    'The shape of prior background_knowledge must be (n_features, n_features)')
 
         # Center variables for each group
         X = X - np.tile(np.mean(X, axis=0), (X.shape[0], 1))
@@ -118,7 +118,7 @@ def fit(self, X):
         return self._estimate_adjacency_matrix(X, prior_knowledge=self._Aknw)
 
     def _extract_partial_orders(self, pk):
-        """ Extract partial orders from prior knowledge."""
+        """ Extract partial orders from prior background_knowledge."""
         path_pairs = np.array(np.where(pk == 1)).transpose()
         no_path_pairs = np.array(np.where(pk == 0)).transpose()
 
@@ -128,7 +128,7 @@ def _extract_partial_orders(self, pk):
             pairs, counts = np.unique(check_pairs, axis=0, return_counts=True)
             if len(pairs[counts > 1]) > 0:
                 raise ValueError(
-                    f'The prior knowledge contains inconsistencies at the following indices: {pairs[counts > 1].tolist()}')
+                    f'The prior background_knowledge contains inconsistencies at the following indices: {pairs[counts > 1].tolist()}')
 
         # Check for inconsistencies in pairs without path
         # If there are duplicate pairs without path, they cancel out and are not ordered.
@@ -141,8 +141,8 @@ def _extract_partial_orders(self, pk):
 
         check_pairs = np.concatenate([path_pairs, no_path_pairs[:, [1, 0]]])
         if len(check_pairs) == 0:
-            # If no pairs are extracted from the specified prior knowledge, 
-            # discard the prior knowledge.
+            # If no pairs are extracted from the specified prior background_knowledge,
+            # discard the prior background_knowledge.
             self._Aknw = None
             return None
 
@@ -151,7 +151,7 @@ def _extract_partial_orders(self, pk):
 
     def _search_candidate(self, U):
         """ Search for candidate features """
-        # If no prior knowledge is specified, nothing to do.
+        # If no prior background_knowledge is specified, nothing to do.
         if self._Aknw is None:
             return U
 
@@ -283,7 +283,7 @@ def _estimate_adjacency_matrix(self, X, prior_knowledge=None):
             Training data, where n_samples is the number of samples
             and n_features is the number of features.
         prior_knowledge : array-like, shape (n_variables, n_variables), optional (default=None)
-            Prior knowledge matrix.
+            Prior background_knowledge matrix.
 
         Returns
         -------
@@ -301,7 +301,7 @@ def _estimate_adjacency_matrix(self, X, prior_knowledge=None):
             # Flatten the array into one dimension
             predictors = self._flatten(self._causal_order[:i])
 
-            # Exclude variables specified in no_path with prior knowledge
+            # Exclude variables specified in no_path with prior background_knowledge
             if prior_knowledge is not None:
                 predictors = [p for p in predictors if pk[target, p] != 0]
 

causallearn/search/FCMBased/lingam/direct_lingam.py

@@ -29,15 +29,15 @@ def __init__(self, random_state=None, prior_knowledge=None, apply_prior_knowledg
         random_state : int, optional (default=None)
             ``random_state`` is the seed used by the random number generator.
         prior_knowledge : array-like, shape (n_features, n_features), optional (default=None)
-            Prior knowledge used for causal discovery, where ``n_features`` is the number of features.
+            Prior background_knowledge used for causal discovery, where ``n_features`` is the number of features.
 
-            The elements of prior knowledge matrix are defined as follows [1]_:
+            The elements of prior background_knowledge matrix are defined as follows [1]_:
 
             * ``0`` : :math:`x_i` does not have a directed path to :math:`x_j`
             * ``1`` : :math:`x_i` has a directed path to :math:`x_j`
-            * ``-1`` : No prior knowledge is available to know if either of the two cases above (0 or 1) is true.
+            * ``-1`` : No prior background_knowledge is available to know if either of the two cases above (0 or 1) is true.
         apply_prior_knowledge_softly : boolean, optional (default=False)
-            If True, apply prior knowledge softly.
+            If True, apply prior background_knowledge softly.
         measure : {'pwling', 'kernel'}, optional (default='pwling')
             Measure to evaluate independence: 'pwling' [2]_ or 'kernel' [1]_.
         """
@@ -50,7 +50,7 @@ def __init__(self, random_state=None, prior_knowledge=None, apply_prior_knowledg
             self._Aknw = check_array(self._Aknw)
             self._Aknw = np.where(self._Aknw < 0, np.nan, self._Aknw)
 
-            # Extract all partial orders in prior knowledge matrix
+            # Extract all partial orders in prior background_knowledge matrix
             if not self._apply_prior_knowledge_softly:
                 self._partial_orders = self._extract_partial_orders(self._Aknw)
 
@@ -74,7 +74,7 @@ def fit(self, X):
         if self._Aknw is not None:
             if (n_features, n_features) != self._Aknw.shape:
                 raise ValueError(
-                    'The shape of prior knowledge must be (n_features, n_features)')
+                    'The shape of prior background_knowledge must be (n_features, n_features)')
 
         # Causal discovery
         U = np.arange(n_features)
@@ -101,7 +101,7 @@ def fit(self, X):
         return self._estimate_adjacency_matrix(X, prior_knowledge=self._Aknw)
 
     def _extract_partial_orders(self, pk):
-        """ Extract partial orders from prior knowledge."""
+        """ Extract partial orders from prior background_knowledge."""
         path_pairs = np.array(np.where(pk == 1)).transpose()
         no_path_pairs = np.array(np.where(pk == 0)).transpose()
 
@@ -111,7 +111,7 @@ def _extract_partial_orders(self, pk):
             pairs, counts = np.unique(check_pairs, axis=0, return_counts=True)
             if len(pairs[counts > 1]) > 0:
                 raise ValueError(
-                    f'The prior knowledge contains inconsistencies at the following indices: {pairs[counts > 1].tolist()}')
+                    f'The prior background_knowledge contains inconsistencies at the following indices: {pairs[counts > 1].tolist()}')
 
         # Check for inconsistencies in pairs without path.
         # If there are duplicate pairs without path, they cancel out and are not ordered.
@@ -124,8 +124,8 @@ def _extract_partial_orders(self, pk):
 
         check_pairs = np.concatenate([path_pairs, no_path_pairs[:, [1, 0]]])
         if len(check_pairs) == 0:
-            # If no pairs are extracted from the specified prior knowledge, 
-            # discard the prior knowledge.
+            # If no pairs are extracted from the specified prior background_knowledge,
+            # discard the prior background_knowledge.
             self._Aknw = None
             return None
 
@@ -152,11 +152,11 @@ def _diff_mutual_info(self, xi_std, xj_std, ri_j, rj_i):
 
     def _search_candidate(self, U):
         """ Search for candidate features """
-        # If no prior knowledge is specified, nothing to do.
+        # If no prior background_knowledge is specified, nothing to do.
         if self._Aknw is None:
             return U, []
 
-        # Apply prior knowledge in a strong way
+        # Apply prior background_knowledge in a strong way
         if not self._apply_prior_knowledge_softly:
             Uc = [i for i in U if i not in self._partial_orders[:, 1]]
             return Uc, []

causallearn/search/FCMBased/lingam/multi_group_direct_lingam.py

@@ -31,15 +31,15 @@ def __init__(self, random_state=None, prior_knowledge=None, apply_prior_knowledg
         random_state : int, optional (default=None)
             ``random_state`` is the seed used by the random number generator.
         prior_knowledge : array-like, shape (n_features, n_features), optional (default=None)
-            Prior knowledge used for causal discovery, where ``n_features`` is the number of features.
+            Prior background_knowledge used for causal discovery, where ``n_features`` is the number of features.
 
-            The elements of prior knowledge matrix are defined as follows [1]_:
+            The elements of prior background_knowledge matrix are defined as follows [1]_:
 
             * ``0`` : :math:`x_i` does not have a directed path to :math:`x_j`
             * ``1`` : :math:`x_i` has a directed path to :math:`x_j`
-            * ``-1`` : No prior knowledge is available to know if either of the two cases above (0 or 1) is true.
+            * ``-1`` : No prior background_knowledge is available to know if either of the two cases above (0 or 1) is true.
         apply_prior_knowledge_softly : boolean, optional (default=False)
-            If True, apply prior knowledge softly.
+            If True, apply prior background_knowledge softly.
         """
         super().__init__(random_state, prior_knowledge, apply_prior_knowledge_softly)
 
@@ -64,7 +64,7 @@ def fit(self, X_list):
         if self._Aknw is not None:
             if (self._n_features, self._n_features) != self._Aknw.shape:
                 raise ValueError(
-                    'The shape of prior knowledge must be (n_features, n_features)')
+                    'The shape of prior background_knowledge must be (n_features, n_features)')
 
         # Causal discovery
         U = np.arange(self._n_features)

causallearn/search/FCMBased/lingam/utils/__init__.py

@@ -61,29 +61,29 @@ def print_dagc(dagc, n_sampling, labels=None):
 
 
 def make_prior_knowledge(n_variables, exogenous_variables=None, sink_variables=None, paths=None, no_paths=None):
-    """ Make matrix of prior knowledge.
+    """ Make matrix of prior background_knowledge.
 
     Parameters
     ----------
     n_variables : int
         Number of variables.
     exogenous_variables : array-like, shape (index, ...), optional (default=None)
         List of exogenous variables(index).
-        Prior knowledge is created with the specified variables as exogenous variables.
+        Prior background_knowledge is created with the specified variables as exogenous variables.
     sink_variables : array-like, shape (index, ...), optional (default=None)
         List of sink variables(index).
-        Prior knowledge is created with the specified variables as sink variables.
+        Prior background_knowledge is created with the specified variables as sink variables.
     paths : array-like, shape ((index, index), ...), optional (default=None)
         List of variables(index) pairs with directed path.
-        If ``(i, j)``, prior knowledge is created that xi has a directed path to xj.
+        If ``(i, j)``, prior background_knowledge is created that xi has a directed path to xj.
     no_paths : array-like, shape ((index, index), ...), optional (default=None)
         List of variables(index) pairs without directed path.
-        If ``(i, j)``, prior knowledge is created that xi does not have a directed path to xj.
+        If ``(i, j)``, prior background_knowledge is created that xi does not have a directed path to xj.
 
     Returns
     -------
     prior_knowledge : array-like, shape (n_variables, n_variables)
-        Return matrix of prior knowledge used for causal discovery.
+        Return matrix of prior background_knowledge used for causal discovery.
     """
     prior_knowledge = np.full((n_variables, n_variables), -1)
     if no_paths:

causallearn/utils/Fas.py

@@ -34,7 +34,7 @@ def forbiddenEdge(node_x, node_y, knowledge):
     if knowledge is None:
         return False
     elif knowledge.is_forbidden(node_x, node_y) and knowledge.is_forbidden(node_y, node_x):
-        print(node_x.get_name() + " --- " + node_y.get_name() + " because it was forbidden by background knowledge.")
+        print(node_x.get_name() + " --- " + node_y.get_name() + " because it was forbidden by background background_knowledge.")
         return True
     return False
 
@@ -148,7 +148,7 @@ def fas(data, nodes, independence_test_method=fisherz, alpha=0.05, knowledge=Non
     node: The search nodes.
     independence_test_method: the independence test method, which should be in causallearn.utils.cit
     alpha: Significance level of independence tests(p_value)(min = 0.00)
-    knowledge: background knowledge
+    knowledge: background background_knowledge
     depth: The depth for the fast adjacency search, or -1 if unlimited
     verbose: True is verbose output should be printed or logged
 
@@ -162,7 +162,7 @@ def fas(data, nodes, independence_test_method=fisherz, alpha=0.05, knowledge=Non
     if (depth is not None) and type(depth) != int:
         raise TypeError("'depth' must be 'int' type!")
     if (knowledge is not None) and type(knowledge) != BackgroundKnowledge:
-        raise TypeError("'knowledge' must be 'BackgroundKnowledge' type!")
+        raise TypeError("'background_knowledge' must be 'BackgroundKnowledge' type!")
 
     # --------end check parameter -----------