Add warning when unable to turn off rounding scheme for a column (#2279)

Author: fealho

sdv/single_table/base.py

@@ -244,8 +244,8 @@ def _validate_transformers(self, column_name_to_transformer):
                     f"Transformer for column '{column}' has already been fit on data."
                 )
 
-    def _warn_for_update_transformers(self, column_name_to_transformer):
-        """Raise warnings for update_transformers.
+    def _warn_quality_and_performance(self, column_name_to_transformer):
+        """Raise warning if the quality/performance may be impacted.
 
         Args:
             column_name_to_transformer (dict):
@@ -259,6 +259,24 @@ def _warn_for_update_transformers(self, column_name_to_transformer):
                     'might impact the quality of your synthetic data.'
                 )
 
+    def _warn_unable_to_enforce_rounding(self, column_name_to_transformer):
+        if self.enforce_rounding:
+            invalid_columns = []
+            for column, transformer in column_name_to_transformer.items():
+                if (
+                    hasattr(transformer, 'learn_rounding_scheme')
+                    and not transformer.learn_rounding_scheme
+                ):
+                    invalid_columns.append(column)
+
+            if invalid_columns:
+                warnings.warn(
+                    f'Unable to turn off rounding scheme for column(s) {invalid_columns}, '
+                    'because the overall synthesizer is enforcing rounding. We '
+                    "recommend setting the synthesizer's 'enforce_rounding' "
+                    'parameter to False.'
+                )
+
     def update_transformers(self, column_name_to_transformer):
         """Update any of the transformers assigned to each of the column names.
 
@@ -267,7 +285,8 @@ def update_transformers(self, column_name_to_transformer):
                 Dict mapping column names to transformers to be used for that column.
         """
         self._validate_transformers(column_name_to_transformer)
-        self._warn_for_update_transformers(column_name_to_transformer)
+        self._warn_quality_and_performance(column_name_to_transformer)
+        self._warn_unable_to_enforce_rounding(column_name_to_transformer)
         self._data_processor.update_transformers(column_name_to_transformer)
         if self._fitted:
             msg = 'For this change to take effect, please refit the synthesizer using `fit`.'

sdv/single_table/copulas.py

@@ -161,8 +161,8 @@ def _fit_model(self, processed_data):
             warnings.filterwarnings('ignore', module='scipy')
             self._model.fit(processed_data)
 
-    def _warn_for_update_transformers(self, column_name_to_transformer):
-        """Raise warnings for update_transformers.
+    def _warn_quality_and_performance(self, column_name_to_transformer):
+        """Raise warning if the quality/performance may be impacted.
 
         Args:
             column_name_to_transformer (dict):

tests/integration/single_table/test_base.py

@@ -844,3 +844,22 @@ def test_fit_int_primary_key_regex_includes_zero(synthesizer_class, regex):
     )
     with pytest.raises(SynthesizerInputError, match=message):
         instance.fit(data)
+
+
+@patch('sdv.single_table.base.warnings')
+def test_update_transformers(warning_mock):
+    """Test the proper warning is raised."""
+    # Setup
+    data, metadata = download_demo(modality='single_table', dataset_name='fake_hotel_guests')
+
+    # Run
+    synthesizer = GaussianCopulaSynthesizer(metadata)
+    synthesizer.auto_assign_transformers(data)
+    synthesizer.update_transformers({'amenities_fee': FloatFormatter(learn_rounding_scheme=False)})
+
+    # Assert
+    warning_mock.warn.assert_called_once_with(
+        "Unable to turn off rounding scheme for column(s) ['amenities_fee'], because the overall "
+        "synthesizer is enforcing rounding. We recommend setting the synthesizer's "
+        "'enforce_rounding' parameter to False."
+    )

tests/unit/single_table/test_base.py

@@ -834,6 +834,31 @@ def test_update_transformers(self):
         assert isinstance(field_transformers['col1'], GaussianNormalizer)
         assert isinstance(field_transformers['col2'], GaussianNormalizer)
 
+    def test_update_transformers_warns_rounding(self):
+        """Test warning is raised if model cannot round."""
+        # Setup
+        column_name_to_transformer = {
+            'col1': GaussianNormalizer(learn_rounding_scheme=False),
+            'col2': GaussianNormalizer(learn_rounding_scheme=True),
+            'col3': GaussianNormalizer(learn_rounding_scheme=False),
+        }
+        metadata = Metadata()
+        instance = BaseSingleTableSynthesizer(metadata)
+        instance._validate_transformers = MagicMock()
+        instance._warn_quality_and_performance = MagicMock()
+        instance._data_processor = MagicMock()
+        instance.enforce_rounding = True
+        instance._fitted = False
+
+        # Run and Assert
+        warn_msg = re.escape(
+            "Unable to turn off rounding scheme for column(s) ['col1', 'col3'], "
+            'because the overall synthesizer is enforcing rounding. We recommend '
+            "setting the synthesizer's 'enforce_rounding' parameter to False."
+        )
+        with pytest.warns(UserWarning, match=warn_msg):
+            instance.update_transformers(column_name_to_transformer)
+
     @patch('sdv.single_table.base.DataProcessor')
     def test__set_random_state(self, mock_data_processor):
         """Test that ``_model.set_random_state`` is being called with the input value.