When running Quality Report, ContingencySimilarity produces a RuntimeWarning (The values in the array are unorderable.)

sdmetrics/column_pairs/statistical/contingency_similarity.py

@@ -44,7 +44,7 @@ def compute(cls, real_data, synthetic_data):
         contingency_synthetic = synthetic.groupby(list(columns), dropna=False).size() / len(
             synthetic
         )
-        combined_index = contingency_real.index.union(contingency_synthetic.index)
+        combined_index = contingency_real.index.union(contingency_synthetic.index, sort=False)
         contingency_synthetic = contingency_synthetic.reindex(combined_index, fill_value=0)
         contingency_real = contingency_real.reindex(combined_index, fill_value=0)
         diff = abs(contingency_real - contingency_synthetic).fillna(0)

tests/integration/reports/multi_table/test_quality_report.py

@@ -342,7 +342,7 @@ def test_quality_report_with_errors():
             None,
         ],
     })
-    assert score == 0.7249603174603174
+    assert score == 0.7249603174603175
     pd.testing.assert_frame_equal(properties, expected_properties)
     pd.testing.assert_frame_equal(details_column_shapes, expected_details)
 

tests/unit/column_pairs/statistical/test_contingency_similarity.py

@@ -1,6 +1,7 @@
 from unittest.mock import patch
 
 import pandas as pd
+import pytest
 
 from sdmetrics.column_pairs.statistical import ContingencySimilarity
 
@@ -53,3 +54,15 @@ def test_normalize(self, normalize_mock):
         # Assert
         normalize_mock.assert_called_once_with(raw_score)
         assert result == normalize_mock.return_value
+
+    @pytest.mark.filterwarnings('error:.*The values in the array are unorderable.*:RuntimeWarning')
+    def test_no_runtime_warning_raised(self):
+        """Test that no RuntimeWarning warning is raised when the metric is computed."""
+        # Setup
+        real_data = pd.DataFrame(data={'A': ['value'] * 4, 'B': ['1', '2', '3', pd.NA]})
+        synthetic_data = pd.DataFrame(data={'A': ['value'] * 3, 'B': ['1', '2', pd.NA]})
+
+        # Run and Assert
+        ContingencySimilarity.compute(
+            real_data=real_data[['A', 'B']], synthetic_data=synthetic_data[['A', 'B']]
+        )