fix: use column names for database mappers (#335)

* fix: use column names for database mappers

When writing results where the metrics include 'confusion_matrix', only
the first column name is written. In the case of the confusion_matrix it
is "true_positive". The desired behaviour is to write all column values.

* Apply same practice for CBPE mapper

* Add tests dealing with result components

---------

Co-authored-by: Niels Nuyttens <niels@nannyml.com>

Author: shezadkhan137

nannyml/io/db/mappers.py

@@ -234,7 +234,9 @@ def _parse(
 
         res: List[DbMetric] = []
 
-        for metric in [metric.column_name for metric in result.metrics]:
+        column_names = [column_name for metric in result.metrics for column_name in metric.column_names]
+
+        for metric in column_names:
             res += (
                 result.filter(partition='analysis')
                 .to_df()[
@@ -288,7 +290,7 @@ def _parse(
 
         res: List[Metric] = []
 
-        for metric in [component[1] for metric in result.metrics for component in metric.components]:
+        for metric in [column_name for metric in result.metrics for column_name in metric.column_names]:
             res += (
                 result.filter(period='analysis')
                 .to_df()[

tests/io/test_writers.py

@@ -112,7 +112,7 @@ def realized_performance_for_binary_classification_result():
         y_true='work_home_actual',
         problem_type='classification_binary',
         timestamp_column_name='timestamp',
-        metrics=['roc_auc', 'f1'],
+        metrics=['roc_auc', 'f1', 'confusion_matrix'],
     ).fit(reference_df)
     result = calc.calculate(analysis_df.merge(analysis_targets_df, on='identifier'))
     return result
@@ -160,7 +160,7 @@ def cbpe_estimated_performance_for_binary_classification_result():
         y_true='work_home_actual',
         problem_type='classification_binary',
         timestamp_column_name='timestamp',
-        metrics=['roc_auc', 'f1'],
+        metrics=['roc_auc', 'f1', 'confusion_matrix'],
     ).fit(reference_df)
     result = calc.estimate(analysis_df.merge(analysis_targets_df, on='identifier'))
     return result
@@ -355,14 +355,14 @@ def test_pickle_file_writer_raises_no_exceptions_when_writing(result):
             'data_reconstruction_feature_drift_metrics',
             10,
         ),
-        (lazy_fixture('realized_performance_for_binary_classification_result'), 'realized_performance_metrics', 40),
+        (lazy_fixture('realized_performance_for_binary_classification_result'), 'realized_performance_metrics', 120),
         (
             lazy_fixture('realized_performance_for_multiclass_classification_result'),
             'realized_performance_metrics',
             40,
         ),
         (lazy_fixture('realized_performance_for_regression_result'), 'realized_performance_metrics', 40),
-        (lazy_fixture('cbpe_estimated_performance_for_binary_classification_result'), 'cbpe_performance_metrics', 20),
+        (lazy_fixture('cbpe_estimated_performance_for_binary_classification_result'), 'cbpe_performance_metrics', 60),
         (
             lazy_fixture('cbpe_estimated_performance_for_multiclass_classification_result'),
             'cbpe_performance_metrics',
@@ -389,3 +389,33 @@ def test_database_writer_exports_correctly(result, table_name, expected_row_coun
 
     finally:
         os.remove('test.db')
+
+
+@pytest.mark.parametrize(
+    'result, table_name',
+    [
+        (lazy_fixture('realized_performance_for_binary_classification_result'), 'realized_performance_metrics'),
+        (lazy_fixture('cbpe_estimated_performance_for_binary_classification_result'), 'cbpe_performance_metrics'),
+    ],
+)
+def test_database_writer_deals_with_metric_components(result, table_name):
+    try:
+        writer = DatabaseWriter(connection_string='sqlite:///test.db', model_name='test')
+        writer.write(result.filter(metrics=['confusion_matrix']))
+
+        import sqlite3
+
+        with sqlite3.connect("test.db", uri=True) as db:
+            res = db.cursor().execute(f"SELECT DISTINCT metric_name FROM {table_name}").fetchall()
+            sut = [row[0] for row in res]
+
+            assert 'true_positive' in sut
+            assert 'false_positive' in sut
+            assert 'true_negative' in sut
+            assert 'false_negative' in sut
+
+    except Exception as exc:
+        pytest.fail(f"an unexpected exception occurred: {exc}")
+
+    finally:
+        os.remove('test.db')