Bug fix for frontdoor identification and a new set of tests (#1093)

* fixed frontdoor bug and added tests

Signed-off-by: Amit Sharma <amit_sharma@live.com>

* updated docstring

Signed-off-by: Amit Sharma <amit_sharma@live.com>

* reformatted file

Signed-off-by: Amit Sharma <amit_sharma@live.com>

---------

Signed-off-by: Amit Sharma <amit_sharma@live.com>

dowhy/causal_estimators/two_stage_regression_estimator.py

@@ -84,6 +84,9 @@ def __init__(
         )
         self.logger.info("INFO: Using Two Stage Regression Estimator")
         # Check if the treatment is one-dimensional
+        if len(self._target_estimand.treatment_variable) > 1:
+            error_msg = str(self.__class__) + "cannot handle more than one treatment variable"
+            raise Exception(error_msg)
         modified_target_estimand = copy.deepcopy(self._target_estimand)
         modified_target_estimand.identifier_method = "backdoor"
         modified_target_estimand.backdoor_variables = self._target_estimand.mediation_first_stage_confounders
@@ -173,23 +176,32 @@ def fit(
         if self._target_estimand.identifier_method == "frontdoor":
             self.logger.debug("Front-door variable used:" + ",".join(self._target_estimand.get_frontdoor_variables()))
             self._frontdoor_variables_names = self._target_estimand.get_frontdoor_variables()
-
             if self._frontdoor_variables_names:
+                if len(self._frontdoor_variables_names) > 1:
+                    raise ValueError(
+                        "Only singleton frontdoor variables are supported for estimation using TwoStageRegression."
+                    )
                 self._frontdoor_variables = data[self._frontdoor_variables_names]
             else:
                 self._frontdoor_variables = None
                 error_msg = "No front-door variable present. Two stage regression is not applicable"
                 self.logger.error(error_msg)
+                raise ValueError(error_msg)
         elif self._target_estimand.identifier_method == "mediation":
             self.logger.debug("Mediators used:" + ",".join(self._target_estimand.get_mediator_variables()))
             self._mediators_names = self._target_estimand.get_mediator_variables()
 
             if self._mediators_names:
+                if len(self._mediators_names) > 1:
+                    raise ValueError(
+                        "Only singleton mediator variables are supported for estimation using TwoStageRegression."
+                    )
                 self._mediators = data[self._mediators_names]
             else:
                 self._mediators = None
                 error_msg = "No mediator variable present. Two stage regression is not applicable"
                 self.logger.error(error_msg)
+                raise ValueError(error_msg)
         elif self._target_estimand.identifier_method == "iv":
             self.logger.debug(
                 "Instrumental variables used:" + ",".join(self._target_estimand.get_instrumental_variables())

dowhy/causal_graph.py

@@ -85,7 +85,7 @@ def __init__(
                     import pydot
 
                     P_list = pydot.graph_from_dot_data(graph)
-                    self._graph = nx.drawing.nx_pydot.from_pydot(P_list[0])
+                    self._graph = nx.DiGraph(nx.drawing.nx_pydot.from_pydot(P_list[0]))
                 except Exception as e:
                     self.logger.error("Error: Pydot cannot be loaded. " + str(e))
                     raise e

dowhy/causal_identifier/auto_identifier.py

@@ -271,7 +271,7 @@ def identify_ate_effect(
 
     ### 3. FRONTDOOR IDENTIFICATION
     # Now checking if there is a valid frontdoor variable
-    frontdoor_variables_names = identify_frontdoor(graph, action_nodes, outcome_nodes)
+    frontdoor_variables_names = identify_frontdoor(graph, action_nodes, outcome_nodes, observed_nodes)
     logger.info("Frontdoor variables for treatment and outcome:" + str(frontdoor_variables_names))
     if len(frontdoor_variables_names) > 0:
         frontdoor_estimand_expr = construct_frontdoor_estimand(
@@ -777,18 +777,18 @@ def build_backdoor_estimands_dict(
 
 
 def identify_frontdoor(
-    graph: nx.DiGraph, action_nodes: List[str], outcome_nodes: List[str], dseparation_algo: str = "default"
+    graph: nx.DiGraph,
+    action_nodes: List[str],
+    outcome_nodes: List[str],
+    observed_nodes: List[str],
+    dseparation_algo: str = "default",
 ):
-    """Find a valid frontdoor variable if it exists.
-
-    Currently only supports a single variable frontdoor set.
-    """
+    """Find a valid frontdoor variable set if it exists."""
     frontdoor_var = None
     frontdoor_paths = None
     fdoor_graph = None
     if dseparation_algo == "default":
         cond1_graph = do_surgery(graph, action_nodes, remove_incoming_edges=True)
-        bdoor_graph1 = do_surgery(graph, action_nodes, remove_outgoing_edges=True)
     elif dseparation_algo == "naive":
         frontdoor_paths = get_all_directed_paths(graph, action_nodes, outcome_nodes)
     else:
@@ -797,48 +797,50 @@ def identify_frontdoor(
     eligible_variables = (
         get_descendants(graph, action_nodes) - set(outcome_nodes) - set(get_descendants(graph, outcome_nodes))
     )
-    # For simplicity, assuming a one-variable frontdoor set
-    for candidate_var in eligible_variables:
-        # Cond 1: All directed paths intercepted by candidate_var
-        cond1 = check_valid_frontdoor_set(
-            graph,
-            action_nodes,
-            outcome_nodes,
-            parse_state(candidate_var),
-            frontdoor_paths=frontdoor_paths,
-            new_graph=cond1_graph,
-            dseparation_algo=dseparation_algo,
-        )
-        logger.debug("Candidate frontdoor set: {0}, is_dseparated: {1}".format(candidate_var, cond1))
-        if not cond1:
-            continue
-        # Cond 2: No confounding between treatment and candidate var
-        cond2 = check_valid_backdoor_set(
-            graph,
-            action_nodes,
-            parse_state(candidate_var),
-            set(),
-            backdoor_paths=None,
-            new_graph=bdoor_graph1,
-            dseparation_algo=dseparation_algo,
-        )
-        if not cond2:
-            continue
-        # Cond 3: treatment blocks all confounding between candidate_var and outcome
-        bdoor_graph2 = do_surgery(graph, candidate_var, remove_outgoing_edges=True)
-        cond3 = check_valid_backdoor_set(
-            graph,
-            parse_state(candidate_var),
-            outcome_nodes,
-            action_nodes,
-            backdoor_paths=None,
-            new_graph=bdoor_graph2,
-            dseparation_algo=dseparation_algo,
-        )
-        is_valid_frontdoor = cond1 and cond2 and cond3
-        if is_valid_frontdoor:
-            frontdoor_var = candidate_var
-            break
+    eligible_variables = eligible_variables.intersection(set(observed_nodes))
+    set_sizes = range(1, len(eligible_variables) + 1, 1)
+    for size_candidate_set in set_sizes:
+        for candidate_set in itertools.combinations(eligible_variables, size_candidate_set):
+            candidate_set = list(candidate_set)
+            # Cond 1: All directed paths intercepted by candidate_var
+            cond1 = check_valid_frontdoor_set(
+                graph,
+                action_nodes,
+                outcome_nodes,
+                candidate_set,
+                frontdoor_paths=frontdoor_paths,
+                new_graph=cond1_graph,
+                dseparation_algo=dseparation_algo,
+            )
+            logger.debug("Candidate frontdoor set: {0}, Cond1: is_dseparated: {1}".format(candidate_set, cond1))
+            if not cond1:
+                continue
+            # Cond 2: No confounding between treatment and candidate var
+            cond2 = check_valid_backdoor_set(
+                graph,
+                action_nodes,
+                candidate_set,
+                set(),
+                backdoor_paths=None,
+                dseparation_algo=dseparation_algo,
+            )["is_dseparated"]
+            if not cond2:
+                continue
+            # Cond 3: treatment blocks all confounding between candidate_var and outcome
+            bdoor_graph2 = do_surgery(graph, candidate_set, remove_outgoing_edges=True)
+            cond3 = check_valid_backdoor_set(
+                graph,
+                candidate_set,
+                outcome_nodes,
+                action_nodes,
+                backdoor_paths=None,
+                new_graph=bdoor_graph2,
+                dseparation_algo=dseparation_algo,
+            )["is_dseparated"]
+            is_valid_frontdoor = cond1 and cond2 and cond3
+            if is_valid_frontdoor:
+                frontdoor_var = candidate_set
+                break
     return parse_state(frontdoor_var)
 
 

dowhy/graph.py

@@ -94,6 +94,7 @@ def check_valid_backdoor_set(
         if new_graph is None:
             # Assume that nodes1 is the treatment
             new_graph = do_surgery(graph, nodes1, remove_outgoing_edges=True)
+
         dseparated = nx.algorithms.d_separated(new_graph, set(nodes1), set(nodes2), set(nodes3))
     elif dseparation_algo == "naive":
         # ignores new_graph parameter, always uses self._graph
@@ -417,7 +418,7 @@ def build_graph_from_str(graph_str: str) -> nx.DiGraph:
                 import pydot
 
                 P_list = pydot.graph_from_dot_data(graph_str)
-                return nx.drawing.nx_pydot.from_pydot(P_list[0])
+                return nx.DiGraph(nx.drawing.nx_pydot.from_pydot(P_list[0]))
             except Exception as e:
                 _logger.error("Error: Pydot cannot be loaded. " + str(e))
                 raise e

tests/causal_estimators/test_two_stage_regression_estimator.py

@@ -143,3 +143,37 @@ def test_frontdoor_estimator(self):
         # Estimate the effect with front-door
         estimate = model.estimate_effect(identified_estimand=estimand, method_name="frontdoor.two_stage_regression")
         assert estimate.value == pytest.approx(0.45, 0.025)
+
+    @mark.parametrize(
+        [
+            "Estimator",
+            "num_treatments",
+            "num_frontdoor_variables",
+        ],
+        [
+            (
+                TwoStageRegressionEstimator,
+                [2, 1],
+                [1, 2],
+            )
+        ],
+    )
+    def test_frontdoor_num_variables_error(self, Estimator, num_treatments, num_frontdoor_variables):
+        estimator_tester = TestEstimator(error_tolerance=0, Estimator=Estimator, identifier_method="frontdoor")
+        with pytest.raises((ValueError, Exception)):
+            estimator_tester.average_treatment_effect_testsuite(
+                num_common_causes=[1, 1],
+                num_instruments=[0, 0],
+                num_effect_modifiers=[0, 0],
+                num_treatments=num_treatments,
+                num_frontdoor_variables=num_frontdoor_variables,
+                treatment_is_binary=[True],
+                outcome_is_binary=[False],
+                confidence_intervals=[
+                    True,
+                ],
+                test_significance=[
+                    False,
+                ],
+                method_params={"num_simulations": 10, "num_null_simulations": 10},
+            )

tests/causal_identifiers/base.py

@@ -2,7 +2,7 @@
 
 from dowhy.graph import build_graph_from_str
 
-from .example_graphs import TEST_GRAPH_SOLUTIONS
+from .example_graphs import TEST_FRONTDOOR_GRAPH_SOLUTIONS, TEST_GRAPH_SOLUTIONS
 
 
 class IdentificationTestGraphSolution(object):
@@ -27,6 +27,27 @@ def __init__(
         )
 
 
+class IdentificationTestFrontdoorGraphSolution(object):
+    def __init__(
+        self,
+        graph_str,
+        observed_variables,
+        valid_frontdoor_sets,
+        invalid_frontdoor_sets,
+    ):
+        self.graph = build_graph_from_str(graph_str)
+        self.action_nodes = ["X"]
+        self.outcome_nodes = ["Y"]
+        self.observed_nodes = observed_variables
+        self.valid_frontdoor_sets = valid_frontdoor_sets
+        self.invalid_frontdoor_sets = invalid_frontdoor_sets
+
+
 @pytest.fixture(params=TEST_GRAPH_SOLUTIONS.keys())
 def example_graph_solution(request):
     return IdentificationTestGraphSolution(**TEST_GRAPH_SOLUTIONS[request.param])
+
+
+@pytest.fixture(params=TEST_FRONTDOOR_GRAPH_SOLUTIONS.keys())
+def example_frontdoor_graph_solution(request):
+    return IdentificationTestFrontdoorGraphSolution(**TEST_FRONTDOOR_GRAPH_SOLUTIONS[request.param])

tests/causal_identifiers/example_graphs.py

@@ -387,3 +387,39 @@
         direct_maximal_adjustment_sets=[{"W1", "M", "W2"}],
     ),
 }
+
+
+TEST_FRONTDOOR_GRAPH_SOLUTIONS = {
+    "valid_singleton": dict(
+        graph_str="digraph {X; Y; Z; M1; X->M1; Z->X; Z->Y; M1->Y;}",
+        observed_variables=["X", "Y", "M1"],
+        valid_frontdoor_sets=[{"M1"}],
+        invalid_frontdoor_sets=[{"Z"}],
+    ),
+    "valid_doubleton": dict(
+        graph_str="digraph {X; Y; Z; M1; M2; X->M1; X->M2; Z->X; Z->Y; M1->Y; M2->Y}",
+        observed_variables=["X", "Y", "M1", "M2"],
+        valid_frontdoor_sets=[
+            {"M1", "M2"},
+        ],
+        invalid_frontdoor_sets=[{"Z"}, {"M1"}, {"M2"}],
+    ),
+    "no_frontdoor": dict(
+        graph_str="digraph{E;X;R;M;Y; E->X;E->R;X->M;R->M;M->Y}",
+        observed_variables=["X", "R", "M", "Y"],
+        valid_frontdoor_sets=[],
+        invalid_frontdoor_sets=[{"M"}, {"M", "E"}, {"M", "R"}],
+    ),
+    "no_frontdoor_simple": dict(
+        graph_str="digraph{X;Y;D;B;X->B;D->B;B->Y;D->Y}",
+        observed_variables=["X", "B", "Y"],
+        valid_frontdoor_sets=[],
+        invalid_frontdoor_sets=[{"B"}, {"D"}],
+    ),
+    "no_frontdoor_in_obs": dict(
+        graph_str="digraph {X; Y; Z; M1; M2; X->M1; X->M2; Z->X; Z->Y; M1->Y; M2->Y}",
+        observed_variables=["X", "Y", "M1", "Z"],
+        valid_frontdoor_sets=[],
+        invalid_frontdoor_sets=[{"Z"}, {"M1"}, {"M2"}, {"M1", "M2"}],
+    ),
+}

tests/causal_identifiers/test_frontdoor_identifier.py

@@ -0,0 +1,50 @@
+import numpy as np
+import pandas as pd
+import pytest
+
+from dowhy import CausalModel
+from dowhy.causal_identifier import AutoIdentifier, BackdoorAdjustment
+from dowhy.causal_identifier.auto_identifier import identify_frontdoor
+from dowhy.causal_identifier.identify_effect import EstimandType
+
+from .base import IdentificationTestFrontdoorGraphSolution, example_frontdoor_graph_solution
+
+
+class TestFrontdoorIdentification(object):
+    def test_identify_frontdoor_functional_api(
+        self, example_frontdoor_graph_solution: IdentificationTestFrontdoorGraphSolution
+    ):
+        graph = example_frontdoor_graph_solution.graph
+        expected_sets = example_frontdoor_graph_solution.valid_frontdoor_sets
+        invalid_sets = example_frontdoor_graph_solution.invalid_frontdoor_sets
+        frontdoor_set = identify_frontdoor(
+            graph,
+            observed_nodes=example_frontdoor_graph_solution.observed_nodes,
+            action_nodes=["X"],
+            outcome_nodes=["Y"],
+        )
+
+        assert (
+            (len(frontdoor_set) == 0) and (len(expected_sets) == 0)
+        ) or (  # No adjustments exist and that's expected.
+            set(frontdoor_set) in expected_sets and set(frontdoor_set) not in invalid_sets
+        )
+
+    def test_identify_frontdoor_causal_model(
+        self, example_frontdoor_graph_solution: IdentificationTestFrontdoorGraphSolution
+    ):
+        graph = example_frontdoor_graph_solution.graph
+        expected_sets = example_frontdoor_graph_solution.valid_frontdoor_sets
+        invalid_sets = example_frontdoor_graph_solution.invalid_frontdoor_sets
+        observed_nodes = example_frontdoor_graph_solution.observed_nodes
+        # Building the causal model
+        num_samples = 10
+        df = pd.DataFrame(np.random.random((num_samples, len(observed_nodes))), columns=observed_nodes)
+        model = CausalModel(data=df, treatment="X", outcome="Y", graph=graph)
+        estimand = model.identify_effect()
+        frontdoor_set = estimand.frontdoor_variables
+        assert (
+            (len(frontdoor_set) == 0) and (len(expected_sets) == 0)
+        ) or (  # No adjustments exist and that's expected.
+            (set(frontdoor_set) in expected_sets) and (set(frontdoor_set) not in invalid_sets)
+        )