fix: cast np.float64 metrics to float for psycopg2, force labels=[0,1] in confusion matrix

Author: William-Hill

ai_model/complete_ml_pipeline.py

@@ -783,7 +783,7 @@ def assign_alert_level(risk_score):
 print(f"F1-Score:  {math_f1:.4f}")
 
 print("\nConfusion Matrix:")
-cm = confusion_matrix(y_test, y_pred)
+cm = confusion_matrix(y_test, y_pred, labels=[0, 1])
 print(f"                Predicted")
 print(f"              No Pass    Pass")
 print(f"Actual No     {cm[0,0]:6d}    {cm[0,1]:6d}")
@@ -894,7 +894,7 @@ def assign_alert_level(risk_score):
 print(f"F1-Score:  {english_f1:.4f}")
 
 print("\nConfusion Matrix:")
-cm = confusion_matrix(y_test, y_pred)
+cm = confusion_matrix(y_test, y_pred, labels=[0, 1])
 print(f"                Predicted")
 print(f"              No Pass    Pass")
 print(f"Actual No     {cm[0,0]:6d}    {cm[0,1]:6d}")
@@ -1008,7 +1008,7 @@ def assign_alert_level(risk_score):
 print(f"F1-Score:  {gpa_f1:.4f}")
 
 print("\nConfusion Matrix:")
-cm = confusion_matrix(y_test, y_pred)
+cm = confusion_matrix(y_test, y_pred, labels=[0, 1])
 print(f"                Predicted")
 print(f"              GPA>=2.0  GPA<2.0")
 print(f"Actual >=2.0  {cm[0,0]:6d}    {cm[0,1]:6d}")

operations/db_utils.py

@@ -191,17 +191,20 @@ def save_model_performance(model_name, model_type, metrics, notes=""):
         VALUES (%s, %s, %s, %s, %s, %s, %s, %s, %s, %s, %s)
         """
 
+        def _f(v):
+            return float(v) if v is not None else None
+
         values = (
             model_name,
             model_type,
-            metrics.get('accuracy'),
-            metrics.get('precision'),
-            metrics.get('recall'),
-            metrics.get('f1'),
-            metrics.get('auc_roc'),
-            metrics.get('rmse'),
-            metrics.get('mae'),
-            metrics.get('r2_score'),
+            _f(metrics.get('accuracy')),
+            _f(metrics.get('precision')),
+            _f(metrics.get('recall')),
+            _f(metrics.get('f1')),
+            _f(metrics.get('auc_roc')),
+            _f(metrics.get('rmse')),
+            _f(metrics.get('mae')),
+            _f(metrics.get('r2_score')),
             notes
         )