tests

Author: R-Palazzo

sdmetrics/single_table/privacy/util.py

@@ -69,9 +69,9 @@ def hamming_distance(target, test):
             The hamming distance
     """
     dist = 0
-    assert len(target) == len(test), (
-        'Tuples must have the same length in the calculation of hamming distance!'
-    )
+    assert len(target) == len(
+        test
+    ), 'Tuples must have the same length in the calculation of hamming distance!'
 
     for target_entry, test_entry in zip(target, test):
         if target_entry != test_entry:

tests/integration/reports/single_table/test_quality_report.py

@@ -77,6 +77,7 @@ def test_report_end_to_end(self):
             key: val for key, val in metadata['columns'].items() if key in column_names
         }
         report = QualityReport()
+        report.num_rows_subsample = None
 
         # Run
         generate_start_time = time.time()
@@ -141,7 +142,8 @@ def test_report_end_to_end(self):
             report.get_details('Column Pair Trends'), expected_details_cpt
         )
         assert report.get_score() == 0.8393750143888287
-
+        assert report._properties['Column Shapes'].num_rows_subsample is None
+        assert report._properties['Column Pair Trends'].num_rows_subsample == None
         report_info = report.get_info()
         assert report_info == report.report_info
 
@@ -183,6 +185,8 @@ def test_with_large_dataset(self):
         # Assert
         cpt_report_1 = report_1.get_properties().iloc[1]['Score']
         cpt_report_2 = report_2.get_properties().iloc[1]['Score']
+        assert report_1._properties['Column Pair Trends'].num_rows_subsample == 50000
+        assert report_2._properties['Column Pair Trends'].num_rows_subsample == 50000
         assert score_1_run_1 != score_1_run_2
         assert np.isclose(score_1_run_1, score_1_run_2, atol=0.001)
         assert np.isclose(report_2.get_score(), score_1_run_1, atol=0.001)

tests/unit/reports/multi_table/test_base_multi_table_report.py

@@ -8,6 +8,7 @@
 
 from sdmetrics.demos import load_demo
 from sdmetrics.reports.multi_table.base_multi_table_report import BaseMultiTableReport
+from sdmetrics.reports.utils import DEFAULT_NUM_ROWS_SUBSAMPLE
 
 
 class TestBaseReport:
@@ -21,6 +22,7 @@ def test__init__(self):
         assert report.is_generated is False
         assert report._properties == {}
         assert report.table_names == []
+        assert report.num_rows_subsample == DEFAULT_NUM_ROWS_SUBSAMPLE
 
     def test__validate_data_format(self):
         """Test the ``_validate_data_format`` method.

tests/unit/reports/test_base_report.py

@@ -9,9 +9,21 @@
 
 from sdmetrics.demos import load_demo
 from sdmetrics.reports.base_report import BaseReport
+from sdmetrics.reports.utils import DEFAULT_NUM_ROWS_SUBSAMPLE
 
 
 class TestBaseReport:
+    def test__init__(self):
+        """Test the initialization of the BaseReport class."""
+        # Run
+        base_report = BaseReport()
+
+        # Assert
+        assert base_report._overall_score is None
+        assert not base_report.is_generated
+        assert base_report._properties == {}
+        assert base_report.num_rows_subsample == DEFAULT_NUM_ROWS_SUBSAMPLE
+
     def test__validate_data_format(self):
         """Test the ``_validate_data_format`` method.
 
@@ -268,6 +280,7 @@ def test_generate(self, version_mock, time_mock, datetime_mock):
         version_mock.return_value = 'version'
 
         base_report = BaseReport()
+        base_report.num_rows_subsample = 1000
         mock_validate = Mock()
         mock__print_results = Mock()
         base_report._print_results = mock__print_results
@@ -292,9 +305,11 @@ def test_generate(self, version_mock, time_mock, datetime_mock):
         base_report._properties['Property 1'].get_score.assert_called_with(
             real_data, synthetic_data, metadata, progress_bar=None
         )
+        assert base_report._properties['Property 1'].num_rows_subsample == 1000
         base_report._properties['Property 2'].get_score.assert_called_with(
             real_data, synthetic_data, metadata, progress_bar=None
         )
+        assert base_report._properties['Property 2'].num_rows_subsample == 1000
         expected_info = {
             'report_type': 'BaseReport',
             'generated_date': '2020-01-05',