Merge pull request #69 from CITCOM-project/remove_intervention

Removed Intervention and all references

Author: Andrew Clark

causal_testing/testing/causal_test_case.py

@@ -1,6 +1,5 @@
 from typing import Any
 
-from causal_testing.testing.intervention import Intervention
 from causal_testing.testing.causal_test_outcome import CausalTestOutcome
 from causal_testing.specification.variable import Variable
 
@@ -18,9 +17,8 @@ class CausalTestCase:
     """
 
     def __init__(self, control_input_configuration: dict[Variable: Any], expected_causal_effect: CausalTestOutcome,
-                 outcome_variables: dict[Variable], intervention: Intervention = None,
-                 treatment_input_configuration: dict[Variable: Any] = None, estimate_type: str = "ate",
-                 effect_modifier_configuration: dict[Variable: Any] = None):
+                 outcome_variables: dict[Variable], treatment_input_configuration: dict[Variable: Any] = None,
+                 estimate_type: str = "ate", effect_modifier_configuration: dict[Variable: Any] = None):
         """
         When a CausalTestCase is initialised, it takes the intervention and applies it to the input configuration to
         create two distinct input configurations: a control input configuration and a treatment input configuration.
@@ -29,35 +27,18 @@ def __init__(self, control_input_configuration: dict[Variable: Any], expected_ca
 
         :param control_input_configuration: The input configuration representing the control values of the treatment
         variables.
-        :param intervention: The function which transforms the control configuration to the treatment configuration.
-        Defaults to None.
         :param treatment_input_configuration: The input configuration representing the treatment values of the treatment
         variables. That is, the input configuration *after* applying the intervention.
         """
-        assert (
-                intervention is None or treatment_input_configuration is None
-        ), 'Cannot define both treatment configuration and intervention.'
-        assert (
-                intervention is not None or treatment_input_configuration is not None
-        ), 'Must define either a treatment configuration or intervention.'
-        if intervention is not None:
-            assert isinstance(intervention, Intervention), \
-                f'Intervention must be an instance of class Intervention not {type(intervention)}.'
-
         self.control_input_configuration = control_input_configuration
         self.expected_causal_effect = expected_causal_effect
-        self.intervention = intervention
         self.outcome_variables = outcome_variables
         self.treatment_input_configuration = treatment_input_configuration
         self.estimate_type = estimate_type
         if effect_modifier_configuration:
             self.effect_modifier_configuration = effect_modifier_configuration
         else:
             self.effect_modifier_configuration = dict()
-        if intervention:
-            self.treatment_input_configuration = intervention.apply(
-                self.control_input_configuration
-            )
         assert self.control_input_configuration.keys() == self.treatment_input_configuration.keys(),\
                "Control and treatment input configurations must have the same keys."
 
@@ -78,12 +59,7 @@ def get_treatment_values(self):
         return list(self.treatment_input_configuration.values())
 
     def __str__(self):
-        if self.intervention is not None:
-            control_input_configuration = {k.name: v for k, v in self.control_input_configuration.items()}
-            return (f"Applying {self.intervention} to {control_input_configuration} should cause the following "
-                    f"changes to {[v.name for v in self.outcome_variables]}: {self.expected_causal_effect}.")
-        else:
-            treatment_config = {k.name: v for k, v in self.treatment_input_configuration.items()}
-            control_config = {k.name: v for k, v in self.control_input_configuration.items()}
-            return (f"Running {treatment_config} instead of {control_config} should cause the following "
-                    f"changes to {self.outcome_variables}: {self.expected_causal_effect}.")
+        treatment_config = {k.name: v for k, v in self.treatment_input_configuration.items()}
+        control_config = {k.name: v for k, v in self.control_input_configuration.items()}
+        return (f"Running {treatment_config} instead of {control_config} should cause the following "
+                f"changes to {self.outcome_variables}: {self.expected_causal_effect}.")

causal_testing/testing/causal_test_engine.py

@@ -32,14 +32,7 @@ class CausalTestEngine:
     def __init__(self, causal_test_case: CausalTestCase, causal_specification: CausalSpecification,
                  data_collector: DataCollector):
         self.causal_test_case = causal_test_case
-        if self.causal_test_case.intervention is not None:
-            self.treatment_variables = list(self.causal_test_case.intervention.treatment_variables)
-        else:
-            # This covers the case where a causal test case is specified in terms of a control/treatment run pair
-            # rather than a control run and an intervention.
-            # It might be better, though, to do this with an intervention which just returns the literal treatment run
-            self.treatment_variables = list(self.causal_test_case.control_input_configuration)
-
+        self.treatment_variables = list(self.causal_test_case.control_input_configuration)
         self.casual_dag, self.scenario = causal_specification.causal_dag, causal_specification.scenario
         self.data_collector = data_collector
         self.scenario_execution_data_df = pd.DataFrame()

examples/covasim_/doubling_beta/causal_test_beta.py

@@ -8,7 +8,6 @@
 from causal_testing.data_collection.data_collector import ObservationalDataCollector
 from causal_testing.testing.causal_test_case import CausalTestCase
 from causal_testing.testing.causal_test_outcome import Positive
-from causal_testing.testing.intervention import Intervention
 from causal_testing.testing.causal_test_engine import CausalTestEngine
 from causal_testing.testing.estimators import LinearRegressionEstimator
 from matplotlib.pyplot import rcParams
@@ -207,8 +206,8 @@ def identification(observational_data_path):
     # 5. Create a causal test case
     causal_test_case = CausalTestCase(control_input_configuration={beta: 0.016},
                                       expected_causal_effect=Positive,
-                                      outcome_variables={cum_infections},
-                                      intervention=Intervention((beta,), (0.032,), ), )
+                                      treatment_input_configuration={beta: 0.032},
+                                      outcome_variables={cum_infections})
 
     # 6. Create a data collector
     data_collector = ObservationalDataCollector(scenario, observational_data_path)

examples/covasim_/vaccinating_elderly/causal_test_vaccine.py

@@ -72,8 +72,8 @@ def experimental_causal_test_vaccinate_elderly(runs_per_test_per_config: int = 3
     for outcome_variable, expected_effect in expected_outcome_effects.items():
         causal_test_case = CausalTestCase(control_input_configuration={vaccine: 0},
                                           expected_causal_effect=expected_effect,
-                                          outcome_variables={outcome_variable},
-                                          intervention=Intervention((vaccine,), (1,), ), )
+                                          treatment_input_configuration={vaccine: 1},
+                                          outcome_variables={outcome_variable})
 
         # 7. Create an instance of the causal test engine
         causal_test_engine = CausalTestEngine(causal_test_case, causal_specification, data_collector)

examples/lr91/causal_test_max_conductances.py

@@ -8,7 +8,6 @@
 from causal_testing.data_collection.data_collector import ObservationalDataCollector
 from causal_testing.testing.causal_test_case import CausalTestCase
 from causal_testing.testing.causal_test_outcome import Positive, Negative, NoEffect
-from causal_testing.testing.intervention import Intervention
 from causal_testing.testing.causal_test_engine import CausalTestEngine
 from causal_testing.testing.estimators import LinearRegressionEstimator
 from matplotlib.pyplot import rcParams
@@ -114,8 +113,8 @@ def effects_on_APD90(observational_data_path, treatment_var, control_val, treatm
     # 6. Create a causal test case
     causal_test_case = CausalTestCase(control_input_configuration={treatment_var: control_val},
                                       expected_causal_effect=expected_causal_effect,
-                                      outcome_variables={apd90},
-                                      intervention=Intervention((treatment_var,), (treatment_val,), ), )
+                                      treatment_input_configuration={treatment_var: treatment_val},
+                                      outcome_variables={apd90})
 
     # 7. Create a data collector
     data_collector = ObservationalDataCollector(scenario, observational_data_path)