feature: extracted analysis config generation for bias post_training

Author: aws-byeldos

src/sagemaker/clarify.py

@@ -1104,16 +1104,13 @@ def run_post_training_bias(
                   the Trial Component will be unassociated.
                 * ``'TrialComponentDisplayName'`` is used for display in Amazon SageMaker Studio.
         """  # noqa E501  # pylint: disable=c0301
-        analysis_config = data_config.get_config()
-        analysis_config.update(data_bias_config.get_config())
-        (
-            probability_threshold,
-            predictor_config,
-        ) = model_predicted_label_config.get_predictor_config()
-        predictor_config.update(model_config.get_predictor_config())
-        analysis_config["methods"] = {"post_training_bias": {"methods": methods}}
-        analysis_config["predictor"] = predictor_config
-        _set(probability_threshold, "probability_threshold", analysis_config)
+        analysis_config = _AnalysisConfigGenerator.bias_post_training(
+            data_config,
+            data_bias_config,
+            model_predicted_label_config,
+            methods,
+            model_config
+        )
         if job_name is None:
             if self.job_name_prefix:
                 job_name = utils.name_from_base(self.job_name_prefix)
@@ -1384,6 +1381,26 @@ def bias_pre_training(data_config, data_bias_config, methods):
         analysis_config["methods"] = {"pre_training_bias": {"methods": methods}}
         return analysis_config
 
+    @staticmethod
+    def bias_post_training(
+            data_config,
+            data_bias_config,
+            model_predicted_label_config,
+            methods,
+            model_config
+    ):
+        analysis_config = data_config.get_config()
+        analysis_config.update(data_bias_config.get_config())
+        analysis_config["methods"] = {"post_training_bias": {"methods": methods}}
+        (
+            probability_threshold,
+            predictor_config,
+        ) = model_predicted_label_config.get_predictor_config()
+        predictor_config.update(model_config.get_predictor_config())
+        analysis_config["predictor"] = predictor_config
+        _set(probability_threshold, "probability_threshold", analysis_config)
+        return analysis_config
+
     @staticmethod
     def _common(analysis_config):
         analysis_config["methods"]["report"] = {

tests/unit/test_clarify.py

@@ -1319,3 +1319,31 @@ def test_analysis_config_generator_for_bias_pre_training(data_config, data_bias_
                 'label_values_or_threshold': [1],
                 'methods': {'pre_training_bias': {'methods': 'all'}}}
     assert actual == expected
+
+
+def test_analysis_config_generator_for_bias_post_training(data_config, data_bias_config, model_config):
+    model_predicted_label_config = ModelPredictedLabelConfig(
+        probability="pr",
+        label_headers=["success"],
+    )
+    actual = _AnalysisConfigGenerator.bias_post_training(
+        data_config,
+        data_bias_config,
+        model_predicted_label_config,
+        methods="all",
+        model_config=model_config,
+    )
+    expected = {'dataset_type': 'text/csv',
+                'facet': [{'name_or_index': 'F1'}],
+                'group_variable': 'F2',
+                'headers': ['Label', 'F1', 'F2', 'F3', 'F4'],
+                'joinsource_name_or_index': 'F4',
+                'label': 'Label',
+                'label_values_or_threshold': [1],
+                'methods': {'post_training_bias': {'methods': 'all'}},
+                'predictor': {'initial_instance_count': 1,
+                              'instance_type': 'ml.c5.xlarge',
+                              'label_headers': ['success'],
+                              'model_name': 'xgboost-model',
+                              'probability': 'pr'}}
+    assert actual == expected