[Repo Assist] improve: replace generic Exception with ValueError in causal estimators

dowhy/causal_estimators/distance_matching_estimator.py

@@ -136,12 +136,12 @@ def fit(self, data: pd.DataFrame, effect_modifier_names: Optional[List[str]] = N
         # 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)
+            raise ValueError(error_msg)
         # Checking if the treatment is binary
         if not data[self._target_estimand.treatment_variable[0]].isin([0, 1]).all():
             error_msg = "Distance Matching method is applicable only for binary treatments."
             self.logger.error(error_msg)
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
 
         self.logger.debug("Adjustment set variables used:" + ",".join(self._target_estimand.get_adjustment_set()))
 
@@ -160,7 +160,7 @@ def fit(self, data: pd.DataFrame, effect_modifier_names: Optional[List[str]] = N
             self._observed_common_causes = None
             error_msg = "No common causes/confounders present. Distance matching methods are not applicable"
             self.logger.error(error_msg)
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
 
         self.symbolic_estimator = self.construct_symbolic_estimator(self._target_estimand)
         self.logger.info(self.symbolic_estimator)

dowhy/causal_estimators/doubly_robust_estimator.py

@@ -122,13 +122,13 @@ def fit(
         # Validate target estimand
         if len(self._target_estimand.treatment_variable) > 1:
             error_msg = str(self.__class__) + " cannot handle more than one treatment variable"
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
         if len(self._target_estimand.outcome_variable) > 1:
             error_msg = str(self.__class__) + " cannot handle more than one outcome variable"
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
         if self._target_estimand.identifier_method not in ["backdoor", "general_adjustment"]:
             error_msg = str(self.__class__) + " only supports covariate adjustment identifiers"
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
         if effect_modifier_names and (len(effect_modifier_names) > 0):
             # TODO: Add support for effect modifiers in the Doubly Robust Estimator
             raise NotImplementedError("Effect Modifiers not supported yet for " + str(self.__class__))

dowhy/causal_estimators/propensity_score_estimator.py

@@ -123,18 +123,18 @@ def fit(
             self._observed_common_causes = None
             error_msg = "No common causes/confounders present. Propensity score based methods are not applicable"
             self.logger.error(error_msg)
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
 
         # 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)
+            raise ValueError(error_msg)
         # Checking if the treatment is binary
         treatment_values = data[self._target_estimand.treatment_variable[0]].astype(int).unique()
         if any([v not in [0, 1] for v in treatment_values]):
             error_msg = "Propensity score methods are applicable only for binary treatments"
             self.logger.error(error_msg)
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
 
         if self.propensity_score_column not in data:
             if self.propensity_score_model is None:

dowhy/causal_estimators/two_stage_regression_estimator.py

@@ -86,7 +86,7 @@ def __init__(
         # 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)
+            raise ValueError(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
@@ -173,7 +173,7 @@ def fit(
 
         if len(self._target_estimand.treatment_variable) > 1:
             error_msg = str(self.__class__) + "cannot handle more than one treatment variable"
-            raise Exception(error_msg)
+            raise ValueError(error_msg)
 
         if self._target_estimand.identifier_method == "frontdoor":
             self.logger.debug("Front-door variable used:" + ",".join(self._target_estimand.get_frontdoor_variables()))

tests/causal_estimators/test_two_stage_regression_estimator.py

@@ -124,9 +124,7 @@ def test_frontdoor_estimator(self):
                 target "X"
             ]
         ]
-        """.replace(
-            "\n", ""
-        )
+        """.replace("\n", "")
 
         N_SAMPLES = 10000
         # Generate the data
@@ -161,7 +159,7 @@ def test_frontdoor_estimator(self):
     )
     def test_frontdoor_num_variables_error(self, Estimator, num_treatments, num_frontdoor_variables):
         estimator_tester = SimpleEstimator(error_tolerance=0, Estimator=Estimator, identifier_method="frontdoor")
-        with pytest.raises((ValueError, Exception)):
+        with pytest.raises(ValueError):
             estimator_tester.average_treatment_effect_testsuite(
                 num_common_causes=[1, 1],
                 num_instruments=[0, 0],
@@ -209,9 +207,7 @@ def _make_mediation_data(n=2000, seed=42):
     edge [ source "X" target "Y" ]
     edge [ source "M" target "Y" ]
 ]
-""".replace(
-    "\n", " "
-)
+""".replace("\n", " ")
 
 
 class TestTwoStageRegressionMediationNIE: