Update app.py

app.py

@@ -118,13 +118,16 @@ def predict_crime_level(crime_felony, crime_misd, crime_viol, sr311_total, dob_p
 
     # Create a DataFrame for the model from all features used during training
     # Ensure the column order matches the training data
-    feature_names = ['crime_felony', 'crime_misd', 'crime_viol', 'sr311_total', 'dob_permits_total']
+    feature_names = None
     
     # Check if the model has feature names attribute
     if hasattr(model, 'feature_name_'):
         feature_names = model.feature_name_
     elif hasattr(model, 'feature_names_in_'):
         feature_names = model.feature_names_in_
+    
+    if feature_names is None:
+        return "Model is missing feature name information.", {}
 
     input_data = pd.DataFrame({
         'crime_felony': [crime_felony],
@@ -135,8 +138,7 @@ def predict_crime_level(crime_felony, crime_misd, crime_viol, sr311_total, dob_p
     })
 
     # Add any other columns from the panel_df that the model might expect, filling with 0
-    all_cols = set(panel_df.columns) - {'GEOID', 'month', 'crime_total'}
-    for col in all_cols:
+    for col in feature_names:
         if col not in input_data.columns:
             input_data[col] = 0
     
@@ -151,7 +153,7 @@ def predict_crime_level(crime_felony, crime_misd, crime_viol, sr311_total, dob_p
     prediction = labels[np.argmax(probabilities)]
     
     # Create a confidence dictionary
-    confidence = {label: prob for label, prob in zip(labels, probabilities)}
+    confidence = {label: f"{prob:.2%}" for label, prob in zip(labels, probabilities)}
     
     return f"Predicted Crime Level: {prediction}", confidence
 
@@ -223,6 +225,7 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
                 map_plot = gr.Plot()
                 ts_plot = gr.Plot()
         
+        # --- Event Listeners ---
         update_button.click(
             fn=create_choropleth_map,
             inputs=[metric_selector, start_date_picker, end_date_picker],
@@ -233,6 +236,18 @@ with gr.Blocks(theme=gr.themes.Soft()) as demo:
             inputs=[metric_selector, start_date_picker, end_date_picker],
             outputs=ts_plot
         )
+        
+        # Trigger initial plot generation on app load
+        demo.load(
+            fn=create_choropleth_map,
+            inputs=[metric_selector, start_date_picker, end_date_picker],
+            outputs=map_plot
+        )
+        demo.load(
+            fn=create_time_series_plot,
+            inputs=[metric_selector, start_date_picker, end_date_picker],
+            outputs=ts_plot
+        )
 
     with gr.Tab("Predictive Analytics"):
         with gr.Tabs():