Update app.py

app.py

@@ -492,8 +492,8 @@ def generate_gpt_response(prompt, dataset):
             st.dataframe(dataset_response)  # Show results to the user
             return f"I found some information in our dataset about {make.title()} {model.title() if model else ''}. Please see the details above."
     
-    # If no match is found, fall back to GPT response
-    openai.api_key = "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"  
+
+    openai.api_key = st.secrets["GPT_TOKEN"]
     system_message = {
         "role": "system",
         "content": (
@@ -698,9 +698,7 @@ def predict_with_ranges(inputs, model, label_encoders):
         'min_price': min_price,
         'max_price': max_price
     }
-# --- Main Application ---
 def main():
-    # Load necessary data and models
     try:
         original_data = load_datasets()
         model, label_encoders = load_model_and_encodings()
@@ -713,23 +711,32 @@ def main():
 
     with tab1:
         st.title("Car Price Prediction")
-        # [Previous prediction interface code]
-        inputs, predict_button = create_prediction_interface()
         
-        if predict_button:
-            st.write(f"Analyzing {inputs['year']} {inputs['make'].title()} {inputs['model'].title()}...")
-            prediction_results = predict_with_ranges(inputs, model, label_encoders)
+        # Create two columns
+        col1, col2 = st.columns([2, 1])
+        
+        with col1:
+            # Prediction interface code
+            inputs, predict_button = create_prediction_interface()
             
-            st.markdown(f"""
-                ### Price Analysis
-                - **Estimated Range**: ${prediction_results['min_price']:,.2f} - ${prediction_results['max_price']:,.2f}
-                - **Model Prediction**: ${prediction_results['predicted_price']:,.2f}
-            """)
-
-            # Generate and display the graph
-            fig = create_market_trends_plot_with_model(model, inputs["make"], inputs, label_encoders)
-            if fig:
-                st.pyplot(fig)
+            if predict_button:
+                st.write(f"Analyzing {inputs['year']} {inputs['make'].title()} {inputs['model'].title()}...")
+                prediction_results = predict_with_ranges(inputs, model, label_encoders)
+                
+                st.markdown(f"""
+                    ### Price Analysis
+                    - **Estimated Range**: ${prediction_results['min_price']:,.2f} - ${prediction_results['max_price']:,.2f}
+                    - **Model Prediction**: ${prediction_results['predicted_price']:,.2f}
+                """)
+
+                # Generate and display the graph
+                fig = create_market_trends_plot_with_model(model, inputs["make"], inputs, label_encoders)
+                if fig:
+                    st.pyplot(fig)
+        
+        with col2:
+            # Add the chat assistant here
+            create_assistant_section(original_data)
 
     with tab2:
         st.title("Car Image Analysis")