feat(simple_metrics): skip threshold alerting below simple_metrics_min_samples (SIENTIAPDE-2189)

laborious/workflows/simple_metrics.py

@@ -33,6 +33,7 @@ class SimpleMetrics:
         target_name = model_config['target']
         model_type = model_config.get('model_type')
         thresholds = model_config.get('simple_metrics_thresholds')
+        min_samples = model_config.get('simple_metrics_min_samples')
 
         # Built from internal Temporal workflow config, not user input (bandit B608).
         query = f"""
@@ -65,6 +66,15 @@ class SimpleMetrics:
         if not target_data:
             return
 
+        # Too few rows for a reliable value: metrics are still persisted, but a noisy short
+        # window must not alert (SIENTIAPDE-1995).
+        enough_samples = not min_samples or len(target_data) >= min_samples
+        if not enough_samples:
+            workflow.logger.info(
+                f'{len(target_data)} rows < simple_metrics_min_samples={min_samples}: '
+                'threshold alerting skipped'
+            )
+
         simple_metrics = await workflow.execute_local_activity_method(
             Activities.calculate_simple_metrics,
             {
@@ -74,7 +84,7 @@ class SimpleMetrics:
                 'metrics': input_data.get('metrics', ['rmse', 'mse', 'mae', 'r2']),
                 'interval_minutes': interval_minutes,
                 'model_type': model_type,
-                'thresholds': thresholds,
+                'thresholds': thresholds if enough_samples else None,
             },
             retry_policy=retry_policy,
             start_to_close_timeout=timedelta(seconds=300),

tests/laborious/workflows/test_simple_metrics.py

@@ -1,4 +1,4 @@
-from unittest.mock import ANY, AsyncMock, call, patch
+from unittest.mock import ANY, AsyncMock, MagicMock, call, patch
 
 from pytest import fixture, mark
 from sientia_do.temporal.constants import DATETIME_FORMAT_WITH_TZ
@@ -217,3 +217,56 @@ async def test_run_default_metrics(workflow_mock: AsyncMock, simple_metrics: Sim
         retry_policy=ANY,
         start_to_close_timeout=ANY,
     )
+
+
+def _input(**model_config) -> dict:
+    return {
+        'schedule_name': 'test_schedule',
+        'model_name': 'test_model',
+        'model_id': 'test_model_id',
+        'interval_minutes': 60,
+        'model_config': {'target': 'test_target', **model_config},
+        'schema': 'test_schema',
+        'predictions_table_name': 'test_predictions_table',
+        'data_table_name': 'test_data_table',
+        'target_table_name': 'test_target_table',
+        'metrics': ['rmse'],
+    }
+
+
+_ROW = {'timestamp': '2026-10-06 10:00:00+0000', 'prediction': 1.0, 'target': 2.0}
+
+
+@mark.asyncio
+@patch('laborious.workflows.simple_metrics.workflow', new_callable=AsyncMock)
+async def test_run_below_min_samples_drops_thresholds(workflow_mock, simple_metrics):
+    workflow_mock.logger = MagicMock()
+    # load, export, and the adaptive-state call S2 (Task 8) adds.
+    workflow_mock.execute_activity_method = AsyncMock(side_effect=[[_ROW] * 2, None, None])
+    workflow_mock.execute_local_activity_method = AsyncMock(
+        return_value=[{'metric': 'rmse', 'value': 9.0}]
+    )
+
+    await simple_metrics.run(
+        _input(simple_metrics_thresholds={'rmse_max': 1.0}, simple_metrics_min_samples=5)
+    )
+
+    activity_input = workflow_mock.execute_local_activity_method.call_args.args[1]
+    assert activity_input['thresholds'] is None
+
+
+@mark.asyncio
+@patch('laborious.workflows.simple_metrics.workflow', new_callable=AsyncMock)
+async def test_run_at_min_samples_keeps_thresholds(workflow_mock, simple_metrics):
+    workflow_mock.logger = MagicMock()
+    workflow_mock.execute_activity_method = AsyncMock(side_effect=[[_ROW] * 5, None, None])
+    workflow_mock.execute_local_activity_method = AsyncMock(
+        return_value=[{'metric': 'rmse', 'value': 9.0}]
+    )
+
+    await simple_metrics.run(
+        _input(simple_metrics_thresholds={'rmse_max': 1.0}, simple_metrics_min_samples=5)
+    )
+
+    activity_input = workflow_mock.execute_local_activity_method.call_args.args[1]
+    assert activity_input['thresholds'] == {'rmse_max': 1.0}