Merge branch 'feat/better-metric-failure-handling'

NannyML · Feb 12, 2024 · a7884ef · a7884ef
2 parents 59fc796 + dead3bd
commit a7884ef
Show file tree

Hide file tree

Showing 4 changed files with 100 additions and 44 deletions.
diff --git a/nannyml/stats/avg/calculator.py b/nannyml/stats/avg/calculator.py
@@ -192,15 +192,29 @@ def _calculate(self, data: pd.DataFrame, *args, **kwargs) -> Result:
 
     def _calculate_for_column(self, data: pd.DataFrame, column_name: str) -> Dict[str, Any]:
         result = {}
-        value = _calculate_avg_value_stats(data[column_name])
-        result['value'] = value
-        result['sampling_error'] = self._sampling_error_components[column_name] / np.sqrt(data.shape[0])
-        result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['upper_threshold'] = self._upper_alert_thresholds[column_name]
-        result['lower_threshold'] = self._lower_alert_thresholds[column_name]
-        result['alert'] = _add_alert_flag(result)
-        return result
+        try:
+            value = _calculate_avg_value_stats(data[column_name])
+            result['value'] = value
+            result['sampling_error'] = self._sampling_error_components[column_name] / np.sqrt(data.shape[0])
+            result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = _add_alert_flag(result)
+        except Exception as exc:
+            if self._logger:
+                self._logger.error(
+                    f"an unexpected exception occurred during calculation of column '{column_name}': " f"{exc}"
+                )
+            result['value'] = np.NaN
+            result['sampling_error'] = np.NaN
+            result['upper_confidence_boundary'] = np.NaN
+            result['lower_confidence_boundary'] = np.NaN
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = np.NaN
+        finally:
+            return result
 
 
 def _create_multilevel_index(

diff --git a/nannyml/stats/median/calculator.py b/nannyml/stats/median/calculator.py
@@ -196,18 +196,32 @@ def _calculate(self, data: pd.DataFrame, *args, **kwargs) -> Result:
 
     def _calculate_for_column(self, data: pd.DataFrame, column_name: str) -> Dict[str, Any]:
         result = {}
-        value = _calculate_median_value_stats(data[column_name])
-        result['value'] = value
-        result['sampling_error'] = summary_stats_median_sampling_error(
-            self._sampling_error_components[column_name], data[column_name]
-        )
-        result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
-
-        result['upper_threshold'] = self._upper_alert_thresholds[column_name]
-        result['lower_threshold'] = self._lower_alert_thresholds[column_name]
-        result['alert'] = _add_alert_flag(result)
-        return result
+        try:
+            value = _calculate_median_value_stats(data[column_name])
+            result['value'] = value
+            result['sampling_error'] = summary_stats_median_sampling_error(
+                self._sampling_error_components[column_name], data[column_name]
+            )
+            result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
+
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = _add_alert_flag(result)
+        except Exception as exc:
+            if self._logger:
+                self._logger.error(
+                    f"an unexpected exception occurred during calculation of column '{column_name}': " f"{exc}"
+                )
+            result['value'] = np.NaN
+            result['sampling_error'] = np.NaN
+            result['upper_confidence_boundary'] = np.NaN
+            result['lower_confidence_boundary'] = np.NaN
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = np.NaN
+        finally:
+            return result
 
 
 def _create_multilevel_index(

diff --git a/nannyml/stats/std/calculator.py b/nannyml/stats/std/calculator.py
@@ -197,21 +197,35 @@ def _calculate(self, data: pd.DataFrame, *args, **kwargs) -> Result:
 
     def _calculate_for_column(self, data: pd.DataFrame, column_name: str) -> Dict[str, Any]:
         result = {}
-        value = _calculate_std_value_stats(data[column_name])
-        result['value'] = value
-        result['sampling_error'] = summary_stats_std_sampling_error(
-            self._sampling_error_components[column_name], data[column_name]
-        )
-        result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['lower_confidence_boundary'] = np.maximum(
-            result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error'],
-            -np.inf if self.lower_threshold_value_limit is None else self.lower_threshold_value_limit
-        )
+        try:
+            value = _calculate_std_value_stats(data[column_name])
+            result['value'] = value
+            result['sampling_error'] = summary_stats_std_sampling_error(
+                self._sampling_error_components[column_name], data[column_name]
+            )
+            result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['lower_confidence_boundary'] = np.maximum(
+                result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error'],
+                -np.inf if self.lower_threshold_value_limit is None else self.lower_threshold_value_limit,
+            )
 
-        result['upper_threshold'] = self._upper_alert_thresholds[column_name]
-        result['lower_threshold'] = self._lower_alert_thresholds[column_name]
-        result['alert'] = _add_alert_flag(result)
-        return result
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = _add_alert_flag(result)
+        except Exception as exc:
+            if self._logger:
+                self._logger.error(
+                    f"an unexpected exception occurred during calculation of column '{column_name}': " f"{exc}"
+                )
+            result['value'] = np.NaN
+            result['sampling_error'] = np.NaN
+            result['upper_confidence_boundary'] = np.NaN
+            result['lower_confidence_boundary'] = np.NaN
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = np.NaN
+        finally:
+            return result
 
 
 def _create_multilevel_index(

diff --git a/nannyml/stats/sum/calculator.py b/nannyml/stats/sum/calculator.py
@@ -191,15 +191,29 @@ def _calculate(self, data: pd.DataFrame, *args, **kwargs) -> Result:
 
     def _calculate_for_column(self, data: pd.DataFrame, column_name: str) -> Dict[str, Any]:
         result = {}
-        value = _calculate_sum_value_stats(data[column_name])
-        result['value'] = value
-        result['sampling_error'] = self._sampling_error_components[column_name] * np.sqrt(data.shape[0])
-        result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
-        result['upper_threshold'] = self._upper_alert_thresholds[column_name]
-        result['lower_threshold'] = self._lower_alert_thresholds[column_name]
-        result['alert'] = _add_alert_flag(result)
-        return result
+        try:
+            value = _calculate_sum_value_stats(data[column_name])
+            result['value'] = value
+            result['sampling_error'] = self._sampling_error_components[column_name] * np.sqrt(data.shape[0])
+            result['upper_confidence_boundary'] = result['value'] + SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['lower_confidence_boundary'] = result['value'] - SAMPLING_ERROR_RANGE * result['sampling_error']
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = _add_alert_flag(result)
+        except Exception as exc:
+            if self._logger:
+                self._logger.error(
+                    f"an unexpected exception occurred during calculation of column '{column_name}': " f"{exc}"
+                )
+            result['value'] = np.NaN
+            result['sampling_error'] = np.NaN
+            result['upper_confidence_boundary'] = np.NaN
+            result['lower_confidence_boundary'] = np.NaN
+            result['upper_threshold'] = self._upper_alert_thresholds[column_name]
+            result['lower_threshold'] = self._lower_alert_thresholds[column_name]
+            result['alert'] = np.NaN
+        finally:
+            return result
 
 
 def _create_multilevel_index(