feature: extracted analysis config generation for bias pre_training

Author: aws-byeldos

src/sagemaker/clarify.py

@@ -1020,9 +1020,11 @@ def run_pre_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())
-        analysis_config["methods"] = {"pre_training_bias": {"methods": methods}}
+        analysis_config = _AnalysisConfigGenerator.bias_pre_training(
+            data_config,
+            data_bias_config,
+            methods
+        )
         if job_name is None:
             if self.job_name_prefix:
                 job_name = utils.name_from_base(self.job_name_prefix)
@@ -1375,6 +1377,21 @@ def explainability(
         analysis_config["predictor"] = predictor_config
         return analysis_config
 
+    @staticmethod
+    def bias_pre_training(data_config, data_bias_config, methods):
+        analysis_config = data_config.get_config()
+        analysis_config.update(data_bias_config.get_config())
+        analysis_config["methods"] = {"pre_training_bias": {"methods": methods}}
+        return analysis_config
+
+    @staticmethod
+    def _common(analysis_config):
+        analysis_config["methods"]["report"] = {
+            "name": "report",
+            "title": "Analysis Report",
+        }
+        return analysis_config
+
 
 def _upload_analysis_config(analysis_config_file, s3_output_path, sagemaker_session, kms_key):
     """Uploads the local ``analysis_config_file`` to the ``s3_output_path``.

tests/unit/test_clarify.py

@@ -1303,3 +1303,19 @@ def test_analysis_config_generator_for_explainability(data_config, model_config)
                               'probability': 'pr'}}
     assert actual == expected
 
+
+def test_analysis_config_generator_for_bias_pre_training(data_config, data_bias_config):
+    actual = _AnalysisConfigGenerator.bias_pre_training(
+        data_config,
+        data_bias_config,
+        methods="all"
+    )
+    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': {'pre_training_bias': {'methods': 'all'}}}
+    assert actual == expected