Update app.py

app.py

@@ -2,101 +2,90 @@ import gradio as gr
 import pandas as pd
 import numpy as np
 import pickle
-from sklearn.preprocessing import LabelEncoder
+from math import radians, cos, sin, asin, sqrt
 
 # Load the model
 with open("apartment_price_model.pkl", mode="rb") as f:
     model = pickle.load(f)
 
-# Load the label encoder for neighborhoods
-with open("neighborhood_encoder.pkl", mode="rb") as f:
-    neighborhood_encoder = pickle.load(f)
+# Zurich neighborhoods with coordinates and distances
+zurich_neighborhoods = {
+    "City Center (Altstadt)": {"lat": 47.3769, "lon": 8.5417, "distance": 0.0},
+    "Oerlikon": {"lat": 47.4111, "lon": 8.5458, "distance": 3.8},
+    "Altstetten": {"lat": 47.3908, "lon": 8.4889, "distance": 4.2},
+    "Wiedikon": {"lat": 47.3708, "lon": 8.5128, "distance": 2.3},
+    "Seefeld": {"lat": 47.3550, "lon": 8.5550, "distance": 2.7},
+    "Schwamendingen": {"lat": 47.4053, "lon": 8.5648, "distance": 3.5},
+    "Wollishofen": {"lat": 47.3517, "lon": 8.5304, "distance": 3.0},
+    "Enge": {"lat": 47.3656, "lon": 8.5267, "distance": 1.2},
+    "Fluntern": {"lat": 47.3797, "lon": 8.5611, "distance": 1.8},
+    "Hottingen": {"lat": 47.3683, "lon": 8.5584, "distance": 1.5},
+    "Custom Location": {"lat": 47.3769, "lon": 8.5417, "distance": 0.0}
+}
 
-def predict_price(neighborhood, rooms, area, has_balcony, is_renovated):
-    # Default values for other features
-    pop = 420217
-    pop_dens = 4778
-    frg_pct = 32.45
-    emp = 491193
-    tax_income = 85446
-    price_per_room = rooms / area if area != 0 else 0
+def predict_price(neighborhood, rooms, area, has_balcony, is_renovated, proximity_to_transport, custom_lat=None, custom_lon=None):
+    if neighborhood == "Custom Location" and custom_lat is not None and custom_lon is not None:
+        lat = custom_lat
+        lon = custom_lon
+    else:
+        lat = zurich_neighborhoods[neighborhood]["lat"]
+        lon = zurich_neighborhoods[neighborhood]["lon"]
 
-    # Encode the neighborhood
-    neighborhood_encoded = neighborhood_encoder.transform([neighborhood])[0]
+    distance_to_center = haversine_distance(lat, lon, 47.3769, 8.5417)
 
-    # Create input dataframe
     input_data = pd.DataFrame([{
         'rooms': rooms,
         'area': area,
-        'pop': pop,
-        'pop_dens': pop_dens,
-        'frg_pct': frg_pct,
-        'emp': emp,
-        'tax_income': tax_income,
-        'price_per_room': price_per_room,
+        'pop': 420217,
+        'pop_dens': 4778,
+        'frg_pct': 32.45,
+        'emp': 491193,
+        'tax_income': 85446,
+        'price_per_room': 0,
+        'distance_to_center': distance_to_center,
         'has_balcony': 1 if has_balcony else 0,
         'is_renovated': 1 if is_renovated else 0,
-        'neighborhood': neighborhood_encoded
+        'proximity_to_transport': 1 if proximity_to_transport else 0
     }])
 
-    # Define features in the correct order
     features = [
         'rooms', 'area', 'pop', 'pop_dens', 'frg_pct', 'emp', 'tax_income',
-        'price_per_room', 'has_balcony', 'is_renovated', 'neighborhood'
+        'price_per_room', 'distance_to_center', 'has_balcony', 'is_renovated', 'proximity_to_transport'
     ]
 
-    # Make prediction
     predicted_price = model.predict(input_data[features])[0]
 
-    # Format the result
     result = f"Predicted Monthly Rent: CHF {predicted_price:.0f}"
     result += f"\n\nProperty Details:"
-    result += f"\n- Neighborhood: {neighborhood}"
+    result += f"\n- Location: {neighborhood}"
     result += f"\n- {rooms} rooms, {area} m²"
+    result += f"\n- {distance_to_center:.2f} km from city center"
     result += f"\n- {'Has balcony' if has_balcony else 'No balcony'}"
     result += f"\n- {'Renovated' if is_renovated else 'Not renovated'}"
-    
-    return result
+    result += f"\n- {'Close to transport' if proximity_to_transport else 'Far from transport'}"
 
-def reset_inputs():
-    return ["City Center (Altstadt)", 3.5, 75, True, False, ""]
+    return result
 
 with gr.Blocks() as demo:
     gr.Markdown("# Zurich Apartment Rent Prediction")
 
     with gr.Row():
         with gr.Column():
-            neighborhood = gr.Dropdown(
-                label="Neighborhood",
-                choices=[
-                    "City Center (Altstadt)", "Oerlikon", "Altstetten", "Wiedikon",
-                    "Seefeld", "Schwamendingen", "Wollishofen", "Enge",
-                    "Fluntern", "Hottingen"
-                ],
-                value="City Center (Altstadt)"
-            )
+            neighborhood = gr.Dropdown(label="Neighborhood", choices=list(zurich_neighborhoods.keys()), value="City Center (Altstadt)")
+            custom_lat = gr.Number(label="Custom Latitude", value=47.3769, visible=False)
+            custom_lon = gr.Number(label="Custom Longitude", value=8.5417, visible=False)
             rooms = gr.Number(label="Number of Rooms", value=3.5)
             area = gr.Number(label="Area (m²)", value=75)
             has_balcony = gr.Checkbox(label="Has Balcony", value=True)
             is_renovated = gr.Checkbox(label="Is Renovated", value=False)
-            
-            with gr.Row():
-                submit_button = gr.Button("Submit")
-                clear_button = gr.Button("Clear")
-
-        with gr.Column():
+            proximity_to_transport = gr.Checkbox(label="Proximity to Transport", value=False)
+            submit_button = gr.Button("Submit")
             output = gr.Textbox(label="Output")
 
-    submit_button.click(
-        fn=predict_price,
-        inputs=[neighborhood, rooms, area, has_balcony, is_renovated],
-        outputs=output
-    )
-
-    clear_button.click(
-        fn=reset_inputs,
-        inputs=None,
-        outputs=[neighborhood, rooms, area, has_balcony, is_renovated, output]
-    )
+        submit_button.click(
+            fn=predict_price,
+            inputs=[neighborhood, rooms, area, has_balcony, is_renovated, proximity_to_transport, custom_lat, custom_lon],
+            outputs=output
+        )
 
 demo.launch()