Add error to NullTransformer when data only contains nans

rdt/errors.py

@@ -10,4 +10,4 @@ class Error(Exception):
 
 
 class TransformerInputError(Exception):
-    """Error to raise when ``HyperTransformer`` receives incorrect input."""
+    """Error to raise when ``HyperTransformer`` receives an incorrect input."""

rdt/transformers/null.py

@@ -5,6 +5,8 @@
 import numpy as np
 import pandas as pd
 
+from rdt.errors import TransformerInputError
+
 LOGGER = logging.getLogger(__name__)
 
 
@@ -52,10 +54,21 @@ def _get_missing_value_replacement(self, data):
         Return:
             object:
                 The fill value that needs to be used.
+
+        Raise:
+            TransformerInputError:
+                Error raised when data only contains nans and ``_missing_value_replacement``
+                is set to 'mean' or  'mode'.
         """
         if self._missing_value_replacement is None:
             return None
 
+        if self._missing_value_replacement in {'mean', 'mode'} and pd.isna(data).all():
+            raise TransformerInputError(
+                f"'missing_value_replacement' cannot be set to '{self._missing_value_replacement}'"
+                ' when the provided data only contains NaNs.'
+            )
+
         if self._missing_value_replacement == 'mean':
             return data.mean()
 

tests/unit/transformers/test_null.py

@@ -1,10 +1,13 @@
 """Unit tests for the NullTransformer."""
 
+import re
 from unittest.mock import patch
 
 import numpy as np
 import pandas as pd
+import pytest
 
+from rdt.errors import TransformerInputError
 from rdt.transformers import NullTransformer
 
 
@@ -98,32 +101,6 @@ def test__get_missing_value_replacement_scalar(self):
         # Assert
         assert missing_value_replacement == 'a_missing_value_replacement'
 
-    def test__get_missing_value_replacement_all_nulls(self):
-        """Test _get_missing_value_replacement when all the values are null.
-
-        If the missing_value_replacement is not a scalar value and all the data
-        values are null, the output be the mean, which is `np.nan`.
-
-        Setup:
-            - NullTransformer passing 'mean' as the missing_value_replacement.
-
-        Input:
-            - A Series filled with nan values.
-            - A np.array of all True values.
-
-        Expected Output:
-            - 0
-        """
-        # Setup
-        transformer = NullTransformer('mean')
-
-        # Run
-        data = pd.Series([np.nan, np.nan, np.nan], name='abc')
-        missing_value_replacement = transformer._get_missing_value_replacement(data)
-
-        # Assert
-        assert missing_value_replacement is np.nan
-
     def test__get_missing_value_replacement_none_numerical(self):
         """Test _get_missing_value_replacement when missing_value_replacement is None.
 
@@ -205,6 +182,20 @@ def test__get_missing_value_replacement_mean(self):
         # Assert
         assert missing_value_replacement == 1.5
 
+    def test__get_missing_value_replacement_mean_only_nans(self):
+        """Test when missing_value_replacement is mean and data only contains nans."""
+        # Setup
+        transformer = NullTransformer('mean')
+        data = pd.Series([float('nan'), None, np.nan], name='abc')
+
+        # Run and Assert
+        err_msg = re.escape(
+            "'missing_value_replacement' cannot be set to 'mean' when "
+            'the provided data only contains NaNs.'
+        )
+        with pytest.raises(TransformerInputError, match=err_msg):
+            transformer._get_missing_value_replacement(data)
+
     def test__get_missing_value_replacement_mode(self):
         """Test _get_missing_value_replacement when missing_value_replacement is 'mode'.
 
@@ -232,6 +223,20 @@ def test__get_missing_value_replacement_mode(self):
         # Assert
         assert missing_value_replacement == 2
 
+    def test__get_missing_value_replacement_mode_only_nans(self):
+        """Test when missing_value_replacement is mode and data only contains nans."""
+        # Setup
+        transformer = NullTransformer('mode')
+        data = pd.Series([float('nan'), None, np.nan], name='abc')
+
+        # Run and Assert
+        err_msg = re.escape(
+            "'missing_value_replacement' cannot be set to 'mode' when "
+            'the provided data only contains NaNs.'
+        )
+        with pytest.raises(TransformerInputError, match=err_msg):
+            transformer._get_missing_value_replacement(data)
+
     def test_fit_model_missing_values_none_and_nulls(self):
         """Test fit when null column is none and there are nulls.