Fix broken tests

Author: frances-h

rdt/transformers/__init__.py

@@ -28,6 +28,7 @@
     ClusterBasedNormalizer,
     FloatFormatter,
     GaussianNormalizer,
+    LogitScaler,
 )
 from rdt.transformers.pii.anonymizer import (
     AnonymizedFaker,

rdt/transformers/numerical.py

@@ -702,9 +702,23 @@ def _fit(self, data):
 
     def _transform(self, data):
         transformed = super()._transform(data)
-        self._validate_logit_inputs(transformed)
-        return logit(transformed, self.min_value, self.max_value)
+        transformed_vals = transformed if transformed.ndim == 1 else transformed[:, 0]
+        self._validate_logit_inputs(transformed_vals)
+        logit_vals = logit(transformed_vals, self.min_value, self.max_value)
+        if transformed.ndim == 1:
+            return logit_vals
+        else:
+            transformed[:, 0] = logit_vals
+            return transformed
 
     def _reverse_transform(self, data):
-        reversed = sigmoid(data, self.min_value, self.max_value)
-        return super()._reverse_transform(reversed)
+        if not isinstance(data, np.ndarray):
+            data = data.to_numpy()
+
+        sampled_vals = data if data.ndim == 1 else data[:, 0]
+        reversed = sigmoid(sampled_vals, self.min_value, self.max_value)
+        if data.ndim == 1:
+            return super()._reverse_transform(reversed)
+        else:
+            data[:, 0] = reversed
+            return super()._reverse_transform(data)

tests/integration/test_transformers.py

@@ -25,10 +25,11 @@
     'GaussianNormalizer': {'missing_value_generation': 'from_column'},
     'ClusterBasedNormalizer': {'missing_value_generation': 'from_column'},
     'LogitScaler': {
+        'missing_value_generation': 'from_column',
         'FROM_DATA': {
-            'min_value': lambda x: np.nanmin(x) - 1,
-            'max_value': lambda x: np.nanmax(x) + 1,
-        }
+            'min_value': lambda x: np.nanmin(x) - 0.01,
+            'max_value': lambda x: np.nanmax(x) + 0.01,
+        },
     },
 }
 

tests/unit/transformers/test_numerical.py

@@ -1961,6 +1961,28 @@ def test__transform(self, mock_logit):
         mock_logit.assert_called_once_with(data, ls.min_value, ls.max_value)
         assert transformed == mock_logit.return_value
 
+    @patch('rdt.transformers.numerical.logit')
+    def test__transform_multi_column(self, mock_logit):
+        """Test the ``transform`` method with multiple columns."""
+        # Setup
+        min_value = (1.0,)
+        max_value = 50.0
+        ls = LogitScaler(min_value=min_value, max_value=max_value)
+        ls._validate_logit_inputs = Mock()
+        data = pd.Series([1.0, 1.1, 1.2, 1.3, 2.0, 3.0, 4.0])
+        null_transformer_mock = Mock()
+        is_null = np.array([0, 0, 0, 1, 0, 1, 0])
+        null_transformer_mock.transform.return_value = np.array([data.to_numpy(), is_null]).T
+        ls.null_transformer = null_transformer_mock
+        logit_values = np.array([0.0, 0.1, 0.2, 0.3, 0.3, 1.4, 2.5])
+        mock_logit.return_value = logit_values
+
+        # Run
+        transformed = ls._transform(data)
+
+        # Assert
+        np.testing.assert_array_equal(transformed, np.array([logit_values, is_null]).T)
+
     @patch('rdt.transformers.numerical.FloatFormatter._reverse_transform')
     @patch('rdt.transformers.numerical.sigmoid')
     def test__reverse_transform(self, mock_sigmoid, ff_reverse_transform_mock):
@@ -1978,6 +2000,37 @@ def test__reverse_transform(self, mock_sigmoid, ff_reverse_transform_mock):
         reversed = ls._reverse_transform(data)
 
         # Assert
-        mock_sigmoid.assert_called_once_with(data, ls.min_value, ls.max_value)
+        mock_sigmoid_args = mock_sigmoid.call_args[0]
+        np.testing.assert_array_equal(mock_sigmoid_args[0], data.to_numpy())
+        assert mock_sigmoid_args[1] == ls.min_value
+        assert mock_sigmoid_args[2] == ls.max_value
         ff_reverse_transform_mock.assert_called_once_with(mock_sigmoid.return_value)
         assert reversed == ff_reverse_transform_mock.return_value
+
+    @patch('rdt.transformers.numerical.FloatFormatter._reverse_transform')
+    @patch('rdt.transformers.numerical.sigmoid')
+    def test__reverse_transform_multi_column(self, mock_sigmoid, ff_reverse_transform_mock):
+        """Test the ``transform`` method with multiple columns."""
+        # Setup
+        min_value = (1.0,)
+        max_value = 50.0
+        ls = LogitScaler(min_value=min_value, max_value=max_value)
+        sampled_data = np.array([1.0, 1.1, 1.2, 1.3, 2.0, 3.0, 4.0])
+        is_null = np.array([0.0, 0.0, 1.0, 1.0, 0.0, 1.0, 1.0])
+        data = pd.DataFrame({'column': sampled_data, 'column.is_null': is_null})
+        null_transformer_mock = Mock()
+        reversed = np.array([1.0, 1.1, np.nan, np.nan, 2.0, np.nan, np.nan])
+        null_transformer_mock.reverse_transform.return_value = reversed
+        ls.null_transformer = null_transformer_mock
+        sigmoid_vals = np.array([3.0, 3.1, 3.3, 3.4, 2.1, 4.0, 4.6])
+        mock_sigmoid.return_value = sigmoid_vals
+
+        # Run
+        reversed = ls._reverse_transform(data)
+
+        # Assert
+        ff_reverse_transform_args = ff_reverse_transform_mock.call_args[0]
+        np.testing.assert_array_equal(
+            ff_reverse_transform_args[0], np.array([sigmoid_vals, is_null]).T
+        )
+        assert reversed == ff_reverse_transform_mock.return_value