Update app.py

app.py

@@ -15,36 +15,36 @@ day_choices = [col.replace('day_of_week_', '') for col in feature_names if col.s
 
 def predict_fall(movement_activity, location, day_of_week, hour_of_day, minute_of_day, time_since_last_event):
     try:
-        # Initialize all features to 0 based on model's feature_names (which must be a list)
-        data = {feat: 0 for feat in feature_names}
+        # 1. Initialize all features to zero using model's expected order
+        data = {f: 0 for f in feature_names}
 
-        # Set categorical one-hot encoded features
+        # 2. Set one-hot encoded categorical features
         data[f'Movement Activity_{movement_activity}'] = 1
         data[f'Location_{location}'] = 1
         data[f'day_of_week_{day_of_week}'] = 1
 
-        # Set numeric features
+        # 3. Set numeric features
         data['hour_of_day'] = hour_of_day
         data['minute_of_day'] = minute_of_day
         data['time_since_last_event'] = time_since_last_event
 
-        # Create input DataFrame in the order of feature_names
+        # 4. Create DataFrame with model feature order
         input_df = pd.DataFrame([data], columns=feature_names)
 
-        print("Scaler expects:", scaler.feature_names_in_)
-        print("Input columns:", input_df.columns.tolist())
+        # 5. Select scaler columns in scaler's order
+        scaler_cols = scaler.feature_names_in_
+        scaler_df = input_df[scaler_cols]
 
-        # Select columns in scaler's expected order
-        input_for_scaler = input_df[scaler.feature_names_in_]
+        # 6. Transform numeric/scaler features
+        scaled_features = scaler.transform(scaler_df)
 
-        # Transform
-        scaled_features = scaler.transform(input_for_scaler)
+        # 7. Replace scaler columns in input_df (which is in model order) with scaled data
+        # Use .loc to avoid dtype warning, cast scaled_features to float explicitly
+        input_df.loc[:, scaler_cols] = scaled_features.astype(float)
 
-        # Replace original columns with scaled data in scaler's order
-        input_df.loc[:, scaler.feature_names_in_] = scaled_features
-
-        # Predict probability
+        # 8. Now predict using the model, passing full input_df in model feature order
         pred_proba = model.predict_proba(input_df)[0, 1]
+
         threshold = 0.4
         pred_label = "Fall Detected" if pred_proba >= threshold else "No Fall"
 
@@ -53,7 +53,7 @@ def predict_fall(movement_activity, location, day_of_week, hour_of_day, minute_o
     except Exception as e:
         import traceback
         traceback.print_exc()
-        return f"Error: {str(e)}. Check logs."
+        return f"Error: {str(e)}. Check server logs."
 
 
 # Gradio interface setup