Update app.py

app.py

@@ -204,21 +204,24 @@ def predict_stock_price(ticker, num_days):
     """
     Main prediction function for Gradio interface
     """
+    # Create empty dataframe for error cases
+    empty_df = pd.DataFrame()
+    
     if not ticker:
-        return None, "Please enter a stock ticker symbol"
+        return None, "Please enter a stock ticker symbol", empty_df
     
     # Convert ticker to uppercase
     ticker = ticker.upper()
     
     # Validate number of days
     if num_days < 1 or num_days > 90:
-        return None, "Please enter a number of days between 1 and 90"
+        return None, "Please enter a number of days between 1 and 90", empty_df
     
     # Get predictions
     historical_data, predictions_df, error = predictor.predict_next_days(ticker, num_days)
     
     if error:
-        return None, error
+        return None, error, empty_df
     
     # Create plot
     fig = predictor.create_plot(historical_data, predictions_df, ticker)
@@ -257,6 +260,78 @@ def predict_stock_price(ticker, num_days):
     
     return fig, summary, predictions_display
 
+# Create demo mode for when models aren't available
+def create_demo_predictions(ticker, num_days):
+    """
+    Create demo predictions when models aren't loaded
+    """
+    # Create fake historical data
+    dates = pd.date_range(end=datetime.now(), periods=100, freq='D')
+    base_price = 150.0
+    historical_data = pd.DataFrame({
+        'price': base_price + np.cumsum(np.random.randn(100) * 2)
+    }, index=dates)
+    
+    # Create fake predictions
+    future_dates = pd.date_range(start=dates[-1] + timedelta(days=1), 
+                                periods=num_days, freq='D')
+    
+    last_price = historical_data['price'].iloc[-1]
+    arima_pred = last_price + np.cumsum(np.random.randn(num_days) * 1.5)
+    lstm_pred = last_price + np.cumsum(np.random.randn(num_days) * 1.5)
+    
+    predictions_df = pd.DataFrame({
+        'Date': future_dates,
+        'ARIMA_Prediction': arima_pred,
+        'LSTM_Prediction': lstm_pred,
+        'Average_Prediction': (arima_pred + lstm_pred) / 2
+    })
+    
+    return historical_data, predictions_df
+
+# Modified predict function with fallback to demo mode
+def predict_stock_price_safe(ticker, num_days):
+    """
+    Safe prediction function with demo fallback
+    """
+    empty_df = pd.DataFrame()
+    
+    if not ticker:
+        return None, "Please enter a stock ticker symbol", empty_df
+    
+    ticker = ticker.upper()
+    
+    if num_days < 1 or num_days > 90:
+        return None, "Please enter a number of days between 1 and 90", empty_df
+    
+    # Check if models are loaded
+    if not all([predictor.arima_model, predictor.lstm_model, predictor.scaler]):
+        # Use demo mode
+        demo_msg = f"""
+        ### ⚠️ Demo Mode Active
+        
+        **Note**: Pre-trained models are not available. Showing demo predictions with random data.
+        
+        To use real predictions, ensure you have:
+        1. `arima_model.pkl` - ARIMA model file
+        2. `lstm_model.h5` - LSTM model file  
+        3. `scaler.pkl` - Data scaler file
+        
+        Place these files in the same directory as the app.
+        """
+        
+        historical_data, predictions_df = create_demo_predictions(ticker, num_days)
+        fig = predictor.create_plot(historical_data, predictions_df, f"{ticker} (DEMO)")
+        
+        predictions_display = predictions_df.copy()
+        predictions_display['Date'] = predictions_display['Date'].dt.strftime('%Y-%m-%d')
+        predictions_display = predictions_display.round(2)
+        
+        return fig, demo_msg, predictions_display
+    
+    # Normal prediction flow
+    return predict_stock_price(ticker, num_days)
+
 # Create Gradio interface
 with gr.Blocks(title="Stock Price Forecaster", theme=gr.themes.Soft()) as app:
     gr.Markdown(
@@ -277,7 +352,7 @@ with gr.Blocks(title="Stock Price Forecaster", theme=gr.themes.Soft()) as app:
         with gr.Column(scale=1):
             ticker_input = gr.Textbox(
                 label="Stock Ticker Symbol",
-                placeholder="e.g., AAPL, GOOGL, MSFT",
+                placeholder="AAPL",
                 value="AAPL"
             )
             
@@ -291,15 +366,7 @@ with gr.Blocks(title="Stock Price Forecaster", theme=gr.themes.Soft()) as app:
             
             predict_button = gr.Button("🚀 Generate Forecast", variant="primary")
             
-            gr.Markdown(
-                """
-                ### Popular Tickers:
-                - **Tech**: AAPL, GOOGL, MSFT, AMZN, TSLA
-                - **Finance**: JPM, BAC, V, MA
-                - **Healthcare**: JNJ, UNH, PFE
-                - **Energy**: XOM, CVX
-                """
-            )
+
     
     with gr.Row():
         with gr.Column(scale=2):
@@ -315,23 +382,20 @@ with gr.Blocks(title="Stock Price Forecaster", theme=gr.themes.Soft()) as app:
             datatype=["str", "number", "number", "number"]
         )
     
-    # Add examples
+    # Add examples (use safe function)
     gr.Examples(
         examples=[
             ["AAPL", 7],
-            ["GOOGL", 14],
-            ["TSLA", 5],
-            ["MSFT", 10]
         ],
         inputs=[ticker_input, days_input],
         outputs=[plot_output, summary_output, predictions_table],
-        fn=predict_stock_price,
+        fn=predict_stock_price_safe,
         cache_examples=False
     )
     
-    # Connect the prediction function
+    # Connect the safe prediction function
     predict_button.click(
-        fn=predict_stock_price,
+        fn=predict_stock_price_safe,
         inputs=[ticker_input, days_input],
         outputs=[plot_output, summary_output, predictions_table]
     )
@@ -343,6 +407,9 @@ with gr.Blocks(title="Stock Price Forecaster", theme=gr.themes.Soft()) as app:
         - **ARIMA**: Auto-Regressive Integrated Moving Average model trained on historical price data
         - **LSTM**: Long Short-Term Memory neural network with 3 layers and dropout regularization
         - **Training Data**: Historical stock prices from Yahoo Finance
+        
+        ### ⚠️ Disclaimer
+        This is for educational purposes only. Stock predictions are inherently uncertain and should not be used as the sole basis for investment decisions.
         """
     )