Add KeyUniqueness metric (#474)

sdmetrics/single_column/__init__.py

@@ -6,6 +6,7 @@
 from sdmetrics.single_column.statistical.category_adherence import CategoryAdherence
 from sdmetrics.single_column.statistical.category_coverage import CategoryCoverage
 from sdmetrics.single_column.statistical.cstest import CSTest
+from sdmetrics.single_column.statistical.key_uniqueness import KeyUniqueness
 from sdmetrics.single_column.statistical.kscomplement import KSComplement
 from sdmetrics.single_column.statistical.missing_value_similarity import MissingValueSimilarity
 from sdmetrics.single_column.statistical.range_coverage import RangeCoverage
@@ -19,6 +20,7 @@
     'CategoryCoverage',
     'CategoryAdherence',
     'CSTest',
+    'KeyUniqueness',
     'KSComplement',
     'MissingValueSimilarity',
     'RangeCoverage',

sdmetrics/single_column/statistical/__init__.py

@@ -4,6 +4,7 @@
 from sdmetrics.single_column.statistical.category_adherence import CategoryAdherence
 from sdmetrics.single_column.statistical.category_coverage import CategoryCoverage
 from sdmetrics.single_column.statistical.cstest import CSTest
+from sdmetrics.single_column.statistical.key_uniqueness import KeyUniqueness
 from sdmetrics.single_column.statistical.kscomplement import KSComplement
 from sdmetrics.single_column.statistical.missing_value_similarity import MissingValueSimilarity
 from sdmetrics.single_column.statistical.range_coverage import RangeCoverage
@@ -15,6 +16,7 @@
     'CategoryCoverage',
     'CategoryAdherence',
     'CSTest',
+    'KeyUniqueness',
     'KSComplement',
     'MissingValueSimilarity',
     'RangeCoverage',

sdmetrics/single_column/statistical/key_uniqueness.py

@@ -0,0 +1,69 @@
+"""Key Uniqueness Metric."""
+import logging
+
+from sdmetrics.goal import Goal
+from sdmetrics.single_column.base import SingleColumnMetric
+
+LOGGER = logging.getLogger(__name__)
+
+
+class KeyUniqueness(SingleColumnMetric):
+    """Key uniqueness metric.
+
+    The proportion of data points in the synthetic data that are unique.
+
+    Attributes:
+        name (str):
+            Name to use when reports about this metric are printed.
+        goal (sdmetrics.goal.Goal):
+            The goal of this metric.
+        min_value (Union[float, tuple[float]]):
+            Minimum value or values that this metric can take.
+        max_value (Union[float, tuple[float]]):
+            Maximum value or values that this metric can take.
+    """
+
+    name = 'KeyUniqueness'
+    goal = Goal.MAXIMIZE
+    min_value = 0.0
+    max_value = 1.0
+
+    @classmethod
+    def compute_breakdown(cls, real_data, synthetic_data):
+        """Compute the score breakdown of the key uniqueness metric.
+
+        Args:
+            real_data (pandas.Series):
+                The real data.
+            synthetic_data (pandas.Series):
+                The synthetic data.
+
+        Returns:
+            dict:
+                The score breakdown of the key uniqueness metric.
+        """
+        has_duplicates = real_data.duplicated().any()
+        has_nans = real_data.isna().any()
+        if has_duplicates or has_nans:
+            LOGGER.info('The real data contains NA or duplicate values.')
+
+        nans_or_duplicates_synthetic = synthetic_data.duplicated() | synthetic_data.isna()
+        score = 1 - nans_or_duplicates_synthetic.sum() / len(synthetic_data)
+
+        return {'score': score}
+
+    @classmethod
+    def compute(cls, real_data, synthetic_data):
+        """Compute the key uniqueness metric.
+
+        Args:
+            real_data (pandas.Series):
+                The real data.
+            synthetic_data (pandas.Series):
+                The synthetic data.
+
+        Returns:
+            float:
+                The proportion of data points in the synthetic data that are unique.
+        """
+        return cls.compute_breakdown(real_data, synthetic_data)['score']

tests/unit/single_column/statistical/test_key_uniqueness.py

@@ -0,0 +1,54 @@
+from unittest.mock import patch
+
+import numpy as np
+import pandas as pd
+
+from sdmetrics.single_column.statistical import KeyUniqueness
+
+
+class TestKeyUniqueness:
+
+    def test_compute_breakdown(self):
+        """Test the ``compute_breakdown`` method."""
+        # Setup
+        real_data = pd.Series([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
+        synthetic_data = pd.Series([1, 2, np.nan, 3, np.nan, 5, 2, np.nan, 6, None])
+
+        metric = KeyUniqueness()
+
+        # Run
+        result = metric.compute_breakdown(real_data, synthetic_data)
+
+        # Assert
+        assert result == {'score': 0.5}
+
+    @patch('sdmetrics.single_column.statistical.key_uniqueness.LOGGER')
+    def test_compute_breakdown_with_duplicates_in_real_data(self, logger_mock):
+        """Test the ``compute_breakdown`` method with duplicates in the real data."""
+        # Setup
+        real_data = pd.Series([1, 2, 3, 4, 5, 6, 7, 8, 9, 10] * 2)
+        synthetic_data = pd.Series([1, 2, np.nan, 3, np.nan, 5, 2, np.nan, 6, None])
+        metric = KeyUniqueness()
+
+        # Run
+        metric.compute_breakdown(real_data, synthetic_data)
+
+        # Assert
+        expected_message = 'The real data contains NA or duplicate values.'
+        logger_mock.info.assert_called_once_with(expected_message)
+
+    @patch('sdmetrics.single_column.statistical.key_uniqueness.KeyUniqueness.compute_breakdown')
+    def test_compute(self, compute_breakdown_mock):
+        """Test the ``compute`` method."""
+        # Setup
+        real_data = pd.Series([1, 2, 3, 4, 5, 6, 7, 8, 9, 10])
+        synthetic_data = pd.Series([1, 2, np.nan, 3, np.nan, 5, 2, np.nan, 6, None])
+        metric = KeyUniqueness()
+        compute_breakdown_mock.return_value = {'score': 0.6}
+
+        # Run
+        result = metric.compute(real_data, synthetic_data)
+
+        # Assert
+        compute_breakdown_mock.assert_called_once_with(real_data, synthetic_data)
+        assert result == 0.6