Update app.py

app.py

@@ -38,12 +38,18 @@ def prepare_predictor():
     return str(predictor_root)
 
 # Load the predictor with version mismatch allowed
-PREDICTOR_DIR = prepare_predictor()
-predictor = TabularPredictor.load(
-    PREDICTOR_DIR, 
-    require_py_version_match=False,
-    require_version_match=False  # Added this parameter to bypass version check
-)
+try:
+    PREDICTOR_DIR = prepare_predictor()
+    predictor = TabularPredictor.load(
+        PREDICTOR_DIR, 
+        require_py_version_match=False,
+        require_version_match=False  # Added this parameter to bypass version check
+    )
+    model_loaded = True
+except Exception as e:
+    print(f"Error loading model: {e}")
+    model_loaded = False
+    predictor = None
 
 # Define feature columns based on the actual model
 FEATURE_COLS = [
@@ -92,64 +98,73 @@ def predict_book_recommendation(
     based on its characteristics and reading status.
     """
     
-    # Encode categorical features
-    fiction_code = FICTION_NONFICTION_MAP[fiction_or_nonfiction]
-    
-    # Create input dataframe with exact feature names
-    input_data = {
-        'FictionorNonfiction': fiction_code,
-        'NumPages': int(num_pages),
-        'ThicknessInches': float(thickness_inches),
-        'ReadUnfinishedorUnread': read_status
-    }
-    
-    df_input = pd.DataFrame([input_data])
+    if not model_loaded or predictor is None:
+        return "⚠️ Model is still loading or failed to load. Please refresh the page.", None
     
-    # Make prediction
-    prediction = predictor.predict(df_input)
-    raw_pred = prediction.iloc[0]
-    
-    # Get prediction probabilities
     try:
-        proba = predictor.predict_proba(df_input)
-        if isinstance(proba, pd.Series):
-            proba = proba.to_frame().T
+        # Encode categorical features
+        fiction_code = FICTION_NONFICTION_MAP[fiction_or_nonfiction]
         
-        # Format probabilities for display
-        proba_dict = {}
-        for cls in proba.columns:
-            label = OUTCOME_LABELS.get(cls, cls)
-            proba_dict[label] = float(proba.iloc[0][cls])
+        # Create input dataframe with exact feature names
+        input_data = {
+            'FictionorNonfiction': fiction_code,
+            'NumPages': int(num_pages),
+            'ThicknessInches': float(thickness_inches),
+            'ReadUnfinishedorUnread': read_status
+        }
         
-        # Sort by probability
-        proba_dict = dict(sorted(proba_dict.items(), key=lambda x: x[1], reverse=True))
-    except Exception as e:
-        print(f"Error getting probabilities: {e}")
-        proba_dict = None
-    
-    # Format prediction label
-    pred_label = OUTCOME_LABELS.get(raw_pred, raw_pred)
-    
-    # Create detailed output
-    output_md = f"## 📚 Prediction Result\n\n"
-    output_md += f"### **{pred_label}**\n\n"
-    
-    if proba_dict:
-        confidence = max(proba_dict.values()) * 100
-        output_md += f"**Confidence:** {confidence:.1f}%\n\n"
-        output_md += "### Probability Distribution:\n"
-        for label, prob in proba_dict.items():
-            bar_length = int(prob * 20)
-            bar = '█' * bar_length + '░' * (20 - bar_length)
-            output_md += f"- {label}: {bar} {prob*100:.1f}%\n"
-    
-    # Add interpretation
-    output_md += "\n### 📊 Input Summary:\n"
-    output_md += f"- **Genre:** {fiction_or_nonfiction}\n"
-    output_md += f"- **Length:** {int(num_pages)} pages ({thickness_inches:.1f} inches thick)\n"
-    output_md += f"- **Reading Status:** {read_status}\n"
+        df_input = pd.DataFrame([input_data])
+        
+        # Make prediction
+        prediction = predictor.predict(df_input)
+        raw_pred = prediction.iloc[0]
+        
+        # Get prediction probabilities
+        try:
+            proba = predictor.predict_proba(df_input)
+            if isinstance(proba, pd.Series):
+                proba = proba.to_frame().T
+            
+            # Format probabilities for display
+            proba_dict = {}
+            for cls in proba.columns:
+                label = OUTCOME_LABELS.get(cls, cls)
+                proba_dict[label] = float(proba.iloc[0][cls])
+            
+            # Sort by probability
+            proba_dict = dict(sorted(proba_dict.items(), key=lambda x: x[1], reverse=True))
+        except Exception as e:
+            print(f"Error getting probabilities: {e}")
+            proba_dict = None
+        
+        # Format prediction label
+        pred_label = OUTCOME_LABELS.get(raw_pred, raw_pred)
+        
+        # Create detailed output
+        output_md = f"## 📚 Prediction Result\n\n"
+        output_md += f"### **{pred_label}**\n\n"
+        
+        if proba_dict:
+            confidence = max(proba_dict.values()) * 100
+            output_md += f"**Confidence:** {confidence:.1f}%\n\n"
+            output_md += "### Probability Distribution:\n"
+            for label, prob in proba_dict.items():
+                bar_length = int(prob * 20)
+                bar = '█' * bar_length + '░' * (20 - bar_length)
+                output_md += f"- {label}: {bar} {prob*100:.1f}%\n"
+        
+        # Add interpretation
+        output_md += "\n### 📊 Input Summary:\n"
+        output_md += f"- **Genre:** {fiction_or_nonfiction}\n"
+        output_md += f"- **Length:** {int(num_pages)} pages ({thickness_inches:.1f} inches thick)\n"
+        output_md += f"- **Reading Status:** {read_status}\n"
+        
+        return output_md, proba_dict
     
-    return output_md, proba_dict
+    except Exception as e:
+        error_msg = f"⚠️ Error making prediction: {str(e)}"
+        print(error_msg)
+        return error_msg, None
 
 # Define example inputs for demonstration
 EXAMPLES = [
@@ -243,18 +258,23 @@ with gr.Blocks(
                 num_top_classes=2
             )
     
-    # Examples section
+    # Examples section - Modified to avoid caching issues
     gr.Markdown("---")
     gr.Markdown("### 💡 Example Books")
     
-    gr.Examples(
-        examples=EXAMPLES,
-        inputs=[fiction_input, pages_input, thickness_input, read_input],
-        outputs=[output_text, output_proba],
-        fn=predict_book_recommendation,
-        label="Click any example to try it:",
-        examples_per_page=6,
-    )
+    try:
+        gr.Examples(
+            examples=EXAMPLES,
+            inputs=[fiction_input, pages_input, thickness_input, read_input],
+            outputs=[output_text, output_proba],
+            fn=predict_book_recommendation,
+            label="Click any example to try it:",
+            examples_per_page=6,
+            cache_examples=False  # Disable caching to avoid startup errors
+        )
+    except Exception as e:
+        print(f"Error creating examples: {e}")
+        gr.Markdown("*Examples could not be loaded*")
     
     # Model information
     gr.Markdown("---")