Changed to XGBoost

app.py

@@ -40,31 +40,60 @@ register_aqi_history_routes(app)
 from aqi_history import get_aqi_history
 print("Initializing AQI History & Model...")
 get_aqi_history()
-from aqi_map import generate_aqi_map_html, render_map_to_png
+from aqi_map import generate_heatmap_html, generate_hotspots_html, render_map_to_png
 import threading
 
 # Ensure map directory exists
 STATIC_DIR = os.path.join(os.getcwd(), 'static')
 os.makedirs(STATIC_DIR, exist_ok=True)
-MAP_IMAGE_PATH = os.path.join(STATIC_DIR, 'aqi_map.png')
+HEATMAP_PATH = os.path.join(STATIC_DIR, 'aqi_map_heatmap.png')
+HOTSPOTS_PATH = os.path.join(STATIC_DIR, 'aqi_map_hotspots.png')
+
+# Default map path (alias to heatmap for backward compat)
+MAP_IMAGE_PATH = HEATMAP_PATH 
 
 @app.route('/api/aqi-map', methods=['GET'])
 def get_aqi_map_html():
-    """Returns the Interactive Folium Map HTML"""
-    return generate_aqi_map_html()
+    """Returns the Interactive Heatmap HTML (Default)"""
+    return generate_heatmap_html()
+
+@app.route('/api/aqi-map/heatmap', methods=['GET'])
+def get_heatmap_html():
+    """Explicit endpoint for Heatmap HTML"""
+    return generate_heatmap_html()
+
+@app.route('/api/aqi-map/hotspots', methods=['GET'])
+def get_hotspots_html():
+    """Returns the Interactive Hotspots HTML"""
+    return generate_hotspots_html()
 
 @app.route('/api/aqi-map.png', methods=['GET'])
 def get_aqi_map_png():
-    """Returns the static PNG image of the map, regenerating if needed"""
+    """Default map image (Heatmap)"""
+    return get_heatmap_png()
+
+@app.route('/api/aqi-map/heatmap.png', methods=['GET'])
+def get_heatmap_png():
+    """Returns the Heatmap Grid PNG"""
+    from flask import send_file
+    
+    # Simple caching/regeneration logic
+    if not os.path.exists(HEATMAP_PATH) or request.args.get('refresh'):
+        html = generate_heatmap_html()
+        render_map_to_png(html, HEATMAP_PATH)
+        
+    return send_file(HEATMAP_PATH, mimetype='image/png')
+
+@app.route('/api/aqi-map/hotspots.png', methods=['GET'])
+def get_hotspots_png():
+    """Returns the Original Hotspots PNG"""
     from flask import send_file
     
-    # Check if we need to regenerate (e.g. if file is old or missing)
-    # For now, we'll generate if missing, or maybe background refresh
-    if not os.path.exists(MAP_IMAGE_PATH):
-        html = generate_aqi_map_html()
-        render_map_to_png(html, MAP_IMAGE_PATH)
+    if not os.path.exists(HOTSPOTS_PATH) or request.args.get('refresh'):
+        html = generate_hotspots_html()
+        render_map_to_png(html, HOTSPOTS_PATH)
         
-    return send_file(MAP_IMAGE_PATH, mimetype='image/png')
+    return send_file(HOTSPOTS_PATH, mimetype='image/png')
 
 # Background task to refresh map periodically (optional)
 def refresh_map_periodically():

aqi_history.py

@@ -4,9 +4,9 @@ import os
 from flask import jsonify, request
 from datetime import datetime, timedelta
 
-# Try importing Random Forest (Standard sklearn)
+# Try importing XGBoost
 try:
-    from sklearn.ensemble import RandomForestRegressor
+    from xgboost import XGBRegressor
     from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
     HAS_MODEL = True
 except ImportError:
@@ -91,7 +91,7 @@ class AQIHistory:
         return df_feat.dropna()
 
     def _train_model(self):
-        print("Training AQI Forecast Model (Random Forest)...")
+        print("Training AQI Forecast Model (XGBoost)...")
         try:
             df_feat = self._create_features(self.df)
             
@@ -105,9 +105,10 @@ class AQIHistory:
             X_train, X_test = X[:split_idx], X[split_idx:]
             y_train, y_test = y[:split_idx], y[split_idx:]
             
-            self.model = RandomForestRegressor(
+            self.model = XGBRegressor(
                 n_estimators=1000,
-                max_depth=20,
+                learning_rate=0.05,
+                max_depth=6,
                 random_state=42,
                 n_jobs=-1
             )
@@ -127,7 +128,7 @@ class AQIHistory:
                 'rmse': float(rmse),
                 'r2': float(r2)
             }
-            print(f"AQI RF Model trained. MAE: {mae:.2f}, R2: {r2:.2f}")
+            print(f"AQI XGB Model trained. MAE: {mae:.2f}, R2: {r2:.2f}")
             
         except Exception as e:
             print(f"Error training AQI model: {e}")

aqi_map.py

@@ -12,266 +12,168 @@ from webdriver_manager.chrome import ChromeDriverManager
 DELHI_LAT = 28.7041
 DELHI_LON = 77.1025
 
-def generate_aqi_map_html():
-    """Generates a Folium map with AQI hotspots and returns the HTML string."""
-    
-    # Create base map
-    m = folium.Map(location=[DELHI_LAT, DELHI_LON], zoom_start=11)
-    
-    # Add Delhi Boundary
+def add_delhi_boundary_to_map(m):
+    """Helper to fetch and add Delhi boundary to a map."""
+    import requests
     try:
-        # Using DataMeet's Delhi Boundary GeoJSON
         geojson_url = "https://raw.githubusercontent.com/datameet/Municipal_Spatial_Data/master/Delhi/Delhi_Boundary.geojson"
+        
+        # Add Boundary Outline
         folium.GeoJson(
             geojson_url,
             name="Delhi Boundary",
             style_function=lambda x: {
                 'fillColor': 'none',
-                'color': 'black',
-                'weight': 2,
-                'dashArray': '5, 5'
+                'color': 'black', # Black for visibility on light tiles
+                'weight': 3,
+                'dashArray': '5, 5',
+                'opacity': 0.6
             }
         ).add_to(m)
+        return True
+    except Exception as e:
+        print(f"Error adding boundary: {e}")
+        return False
+
+def generate_heatmap_html():
+    """Generates a Folium map with Grid Heatmap (Clipped to Delhi)."""
+    
+    # Create base map
+    m = folium.Map(
+        location=[DELHI_LAT, DELHI_LON], 
+        zoom_start=10, 
+        control_scale=True
+    )
+    
+    # Heatmap Grid Generation
+    import requests
+    from shapely.geometry import shape, Point
+    from shapely.ops import unary_union
+
+    # Fetch Delhi Boundary for layout clipping & display
+    # We fetch manually here for the shapely polygon, AND add the visual layer
+    has_boundary = False
+    delhi_polygon = None
+    
+    # Add visual boundary
+    add_delhi_boundary_to_map(m)
+
+    try:
+        geojson_url = "https://raw.githubusercontent.com/datameet/Municipal_Spatial_Data/master/Delhi/Delhi_Boundary.geojson"
+        resp = requests.get(geojson_url)
+        data = resp.json()
+
+        # Create a unified polygon for checking
+        features = data.get('features', [])
+        shapes = [shape(f['geometry']) for f in features]
+        delhi_polygon = unary_union(shapes)
+        has_boundary = True
     except Exception as e:
-        print(f"Error loading Delhi GeoJSON: {e}")
+        print(f"Error processing boundary for clipping: {e}")
+    
+    # Grid config
+    lat_min, lat_max = 28.40, 28.88
+    lon_min, lon_max = 76.85, 77.35
+    step = 0.015  # Approx 1.5km grid size
+    
+    def get_color(value):
+        """Red-scale heatmap colors like the reference image."""
+        if value < 100: return "#fee5d9"  # Very Light Pink
+        if value < 200: return "#fcae91"  # Light Pink/Orange
+        if value < 300: return "#fb6a4a"  # Salmon
+        if value < 400: return "#de2d26"  # Red
+        return "#a50f15"                  # Dark Red/Brown
+
+    lat = lat_min
+    while lat < lat_max:
+        lon = lon_min
+        while lon < lon_max:
+            # Check if center of grid is inside Delhi
+            center_point = Point(lon + step/2, lat + step/2)
+            
+            if has_boundary and not delhi_polygon.contains(center_point):
+                # Skip if outside
+                lon += step
+                continue
+
+            # Generate simulated AQI/CO2 data with spatial coherence
+            # Higher near center (Delhi)
+            dist_center = ((lat - DELHI_LAT)**2 + (lon - DELHI_LON)**2)**0.5
+            
+            # Base value + random noise - distance decay
+            base_value = 450 - (dist_center * 800) 
+            val = base_value + random.randint(-50, 50)
+            val = max(50, min(500, val)) # Clamp 50-500
+            
+            color = get_color(val)
+            
+            # Draw grid cell
+            folium.Rectangle(
+                bounds=[[lat, lon], [lat + step, lon + step]],
+                color=None,        # No border
+                fill=True,
+                fill_color=color,
+                fill_opacity=0.6, # Slightly transparent
+                tooltip=f"Zone AQI: {int(val)}"
+            ).add_to(m)
+            
+            lon += step
+        lat += step
+
+    return m.get_root().render()
+
+def generate_hotspots_html():
+    """Generates the original Hotspot Map with markers."""
+    m = folium.Map(location=[DELHI_LAT, DELHI_LON], zoom_start=11)
+    
+    # Add Visual Boundary
+    add_delhi_boundary_to_map(m)
     
-    # Randomly generate some "hotspot" locations around Delhi for demo
-    # In a real app, this would come from the AQI data source
+    # Original Hotspots Data
     hotspots = [
-        {
-            "name": "Chandni Chowk",
-            "lat": 28.656,
-            "lon": 77.231,
-            "aqi": 384
-        },
-        {
-            "name": "Nehru Nagar",
-            "lat": 28.5697,
-            "lon": 77.253,
-            "aqi": 384
-        },
-        {
-            "name": "New Moti Bagh",
-            "lat": 28.582,
-            "lon": 77.1717,
-            "aqi": 377
-        },
-        {
-            "name": "Shadipur",
-            "lat": 28.6517,
-            "lon": 77.1582,
-            "aqi": 375
-        },
-        {
-            "name": "Wazirpur",
-            "lat": 28.6967,
-            "lon": 77.1658,
-            "aqi": 354
-        },
-        {
-            "name": "Ashok Vihar",
-            "lat": 28.6856,
-            "lon": 77.178,
-            "aqi": 231
-        },
-        {
-            "name": "Mundka",
-            "lat": 28.6794,
-            "lon": 77.0284,
-            "aqi": 353
-        },
-        {
-            "name": "Punjabi Bagh",
-            "lat": 28.673,
-            "lon": 77.1374,
-            "aqi": 349
-        },
-        {
-            "name": "RK Puram",
-            "lat": 28.5505,
-            "lon": 77.1849,
-            "aqi": 342
-        },
-        {
-            "name": "Jahangirpuri",
-            "lat": 28.716,
-            "lon": 77.1829,
-            "aqi": 338
-        },
-        {
-            "name": "Okhla Phase-2",
-            "lat": 28.5365,
-            "lon": 77.2803,
-            "aqi": 337
-        },
-        {
-            "name": "Dwarka Sector-8",
-            "lat": 28.5656,
-            "lon": 77.067,
-            "aqi": 336
-        },
-        {
-            "name": "Patparganj",
-            "lat": 28.612,
-            "lon": 77.292,
-            "aqi": 334
-        },
-        {
-            "name": "Bawana",
-            "lat": 28.8,
-            "lon": 77.03,
-            "aqi": 328
-        },
-        {
-            "name": "Sonia Vihar",
-            "lat": 28.7074,
-            "lon": 77.2599,
-            "aqi": 317
-        },
-        {
-            "name": "Rohini",
-            "lat": 28.7019,
-            "lon": 77.0984,
-            "aqi": 312
-        },
-        {
-            "name": "Narela",
-            "lat": 28.85,
-            "lon": 77.1,
-            "aqi": 311
-        },
-        {
-            "name": "Mandir Marg",
-            "lat": 28.6325,
-            "lon": 77.1994,
-            "aqi": 311
-        },
-        {
-            "name": "Vivek Vihar",
-            "lat": 28.6635,
-            "lon": 77.3152,
-            "aqi": 303
-        },
-        {
-            "name": "Anand Vihar",
-            "lat": 28.6508,
-            "lon": 77.3152,
-            "aqi": 335
-        },
-        {
-            "name": "Dhyanchand Stadium",
-            "lat": 28.6125,
-            "lon": 77.2372,
-            "aqi": 335
-        },
-        {
-            "name": "Sirifort",
-            "lat": 28.5521,
-            "lon": 77.2193,
-            "aqi": 326
-        },
-        {
-            "name": "Karni Singh Shooting Range",
-            "lat": 28.4998,
-            "lon": 77.2668,
-            "aqi": 323
-        },
-        {
-            "name": "ITO",
-            "lat": 28.6294,
-            "lon": 77.241,
-            "aqi": 324
-        },
-        {
-            "name": "JLN Stadium",
-            "lat": 28.5828,
-            "lon": 77.2344,
-            "aqi": 314
-        },
-        {
-            "name": "IGI Airport T3",
-            "lat": 28.5562,
-            "lon": 77.1,
-            "aqi": 305
-        },
-        {
-            "name": "Pusa IMD",
-            "lat": 28.6335,
-            "lon": 77.1651,
-            "aqi": 298
-        },
-        {
-            "name": "Burari Crossing",
-            "lat": 28.7592,
-            "lon": 77.1938,
-            "aqi": 294
-        },
-        {
-            "name": "Aurobindo Marg",
-            "lat": 28.545,
-            "lon": 77.205,
-            "aqi": 287
-        },
-        {
-            "name": "Dilshad Garden (IHBAS)",
-            "lat": 28.681,
-            "lon": 77.305,
-            "aqi": 283
-        },
-        {
-            "name": "Lodhi Road",
-            "lat": 28.5921,
-            "lon": 77.2284,
-            "aqi": 282
-        },
-        {
-            "name": "NSIT-Dwarka",
-            "lat": 28.6081,
-            "lon": 77.0193,
-            "aqi": 275
-        },
-        {
-            "name": "Alipur",
-            "lat": 28.8,
-            "lon": 77.15,
-            "aqi": 273
-        },
-        {
-            "name": "Najafgarh",
-            "lat": 28.6125,
-            "lon": 76.983,
-            "aqi": 262
-        },
-        {
-            "name": "CRRI-Mathura Road",
-            "lat": 28.5518,
-            "lon": 77.2752,
-            "aqi": 252
-        },
-        {
-            "name": "DTU",
-            "lat": 28.7501,
-            "lon": 77.1177,
-            "aqi": 219
-        },
-        {
-            "name": "DU North Campus",
-            "lat": 28.69,
-            "lon": 77.21,
-            "aqi": 298
-        },
-        {
-            "name": "Ayanagar",
-            "lat": 28.4809,
-            "lon": 77.1255,
-            "aqi": 342
-        }
+        {"name": "Chandni Chowk", "lat": 28.656, "lon": 77.231, "aqi": 384},
+        {"name": "Nehru Nagar", "lat": 28.5697, "lon": 77.253, "aqi": 384},
+        {"name": "New Moti Bagh", "lat": 28.582, "lon": 77.1717, "aqi": 377},
+        {"name": "Shadipur", "lat": 28.6517, "lon": 77.1582, "aqi": 375},
+        {"name": "Wazirpur", "lat": 28.6967, "lon": 77.1658, "aqi": 354},
+        {"name": "Ashok Vihar", "lat": 28.6856, "lon": 77.178, "aqi": 231},
+        {"name": "Mundka", "lat": 28.6794, "lon": 77.0284, "aqi": 353},
+        {"name": "Punjabi Bagh", "lat": 28.673, "lon": 77.1374, "aqi": 349},
+        {"name": "RK Puram", "lat": 28.5505, "lon": 77.1849, "aqi": 342},
+        {"name": "Jahangirpuri", "lat": 28.716, "lon": 77.1829, "aqi": 338},
+        {"name": "Okhla Phase-2", "lat": 28.5365, "lon": 77.2803, "aqi": 337},
+        {"name": "Dwarka Sector-8", "lat": 28.5656, "lon": 77.067, "aqi": 336},
+        {"name": "Patparganj", "lat": 28.612, "lon": 77.292, "aqi": 334},
+        {"name": "Bawana", "lat": 28.8, "lon": 77.03, "aqi": 328},
+        {"name": "Sonia Vihar", "lat": 28.7074, "lon": 77.2599, "aqi": 317},
+        {"name": "Rohini", "lat": 28.7019, "lon": 77.0984, "aqi": 312},
+        {"name": "Narela", "lat": 28.85, "lon": 77.1, "aqi": 311},
+        {"name": "Mandir Marg", "lat": 28.6325, "lon": 77.1994, "aqi": 311},
+        {"name": "Vivek Vihar", "lat": 28.6635, "lon": 77.3152, "aqi": 303},
+        {"name": "Anand Vihar", "lat": 28.6508, "lon": 77.3152, "aqi": 335},
+        {"name": "Dhyanchand Stadium", "lat": 28.6125, "lon": 77.2372, "aqi": 335},
+        {"name": "Sirifort", "lat": 28.5521, "lon": 77.2193, "aqi": 326},
+        {"name": "Karni Singh Shooting Range", "lat": 28.4998, "lon": 77.2668, "aqi": 323},
+        {"name": "ITO", "lat": 28.6294, "lon": 77.241, "aqi": 324},
+        {"name": "JLN Stadium", "lat": 28.5828, "lon": 77.2344, "aqi": 314},
+        {"name": "IGI Airport T3", "lat": 28.5562, "lon": 77.1, "aqi": 305},
+        {"name": "Pusa IMD", "lat": 28.6335, "lon": 77.1651, "aqi": 298},
+        {"name": "Burari Crossing", "lat": 28.7592, "lon": 77.1938, "aqi": 294},
+        {"name": "Aurobindo Marg", "lat": 28.545, "lon": 77.205, "aqi": 287},
+        {"name": "Dilshad Garden (IHBAS)", "lat": 28.681, "lon": 77.305, "aqi": 283},
+        {"name": "Lodhi Road", "lat": 28.5921, "lon": 77.2284, "aqi": 282},
+        {"name": "NSIT-Dwarka", "lat": 28.6081, "lon": 77.0193, "aqi": 275},
+        {"name": "Alipur", "lat": 28.8, "lon": 77.15, "aqi": 273},
+        {"name": "Najafgarh", "lat": 28.6125, "lon": 76.983, "aqi": 262},
+        {"name": "CRRI-Mathura Road", "lat": 28.5518, "lon": 77.2752, "aqi": 252},
+        {"name": "DTU", "lat": 28.7501, "lon": 77.1177, "aqi": 219},
+        {"name": "DU North Campus", "lat": 28.69, "lon": 77.21, "aqi": 298},
+        {"name": "Ayanagar", "lat": 28.4809, "lon": 77.1255, "aqi": 342}
     ]
-    
+
     for spot in hotspots:
-        color = "gray" # Default for NA or invalid
+        color = "gray"
         aqi = spot['aqi']
-        
         if isinstance(aqi, (int, float)):
              color = "green"
              if aqi > 100: color = "yellow"
@@ -279,10 +181,9 @@ def generate_aqi_map_html():
              if aqi > 300: color = "red"
              if aqi > 400: color = "purple"
         
-        # Add marker
         folium.CircleMarker(
             location=[spot['lat'], spot['lon']],
-            radius=20,
+            radius=15,
             popup=f"{spot['name']}: AQI {spot['aqi']}",
             color=color,
             fill=True,
@@ -292,7 +193,7 @@ def generate_aqi_map_html():
         
         folium.Marker(
             location=[spot['lat'], spot['lon']],
-            icon=folium.DivIcon(html=f'<div style="font-weight: bold; color: black;">{spot["aqi"]}</div>')
+            icon=folium.DivIcon(html=f'<div style="font-weight: bold; color: white; text-shadow: 0 0 3px black;">{spot["aqi"]}</div>')
         ).add_to(m)
         
     return m.get_root().render()
@@ -301,7 +202,9 @@ def render_map_to_png(html_content, output_path):
     """
     Renders the HTML content to a PNG image using Selenium headless Chrome.
     """
-    tmp_html_path = os.path.abspath("temp_map.html")
+    import uuid
+    unique_id = uuid.uuid4()
+    tmp_html_path = os.path.abspath(f"temp_map_{unique_id}.html")
     
     # Write HTML to temp file
     with open(tmp_html_path, "w", encoding="utf-8") as f:

static/aqi_map.png → aqi_metrics.pkl

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:687c30365da87046c8524385e8098a2db608a45d42a368e9f4788ba97d4b601e
-size 433996
+oid sha256:f32f773d93a81034f6c95c72aca695424d5540bc1cf93800b28c80e122160db5
+size 76

aqi_model.pkl

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:6cc68866cbf743a194980b11eaa30b172647edb1fff701f4d0660c7109178298
+size 191992993

graph_state.json

@@ -0,0 +1,68 @@
+{
+  "nodes": [
+    {"id": "industries", "type": "causal", "position": {"x": 600, "y": 0}, "data": {"label": "Industries", "value": 100, "enabled": true, "type": "sector"}},
+    {"id": "transport", "type": "causal", "position": {"x": 600, "y": 150}, "data": {"label": "Transport", "value": 35, "enabled": true, "type": "sector"}},
+    {"id": "energy", "type": "causal", "position": {"x": 900, "y": 320}, "data": {"label": "Energy", "value": 0, "enabled": true, "type": "sector"}},
+    {"id": "infrastructure", "type": "causal", "position": {"x": 200, "y": 420}, "data": {"label": "Infrastructure", "value": 0, "enabled": true, "type": "sector"}},
+    {"id": "moves-goods", "type": "causal", "position": {"x": 600, "y": 80}, "data": {"label": "Moves goods → ↑ CO₂ & particulates", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "uses-power", "type": "causal", "position": {"x": 850, "y": 120}, "data": {"label": "Uses power → ↑ CO₂ & pollutants", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "powers-industry", "type": "causal", "position": {"x": 1050, "y": 120}, "data": {"label": "Powers industry → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "industrial-pollutants", "type": "causal", "position": {"x": 1150, "y": 320}, "data": {"label": "Industrial pollutants (PM, NOx, SO₂)", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "fuels-transport", "type": "causal", "position": {"x": 750, "y": 240}, "data": {"label": "Fuels transport → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "fuel-refining", "type": "causal", "position": {"x": 950, "y": 240}, "data": {"label": "Fuel refining → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "vehicle-emissions", "type": "causal", "position": {"x": 1150, "y": 450}, "data": {"label": "Vehicle emissions (PM, NOx)", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "urban-sprawl", "type": "causal", "position": {"x": 600, "y": 280}, "data": {"label": "Urban sprawl → ↑ CO₂ & emissions", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "drives-construction", "type": "causal", "position": {"x": 100, "y": 240}, "data": {"label": "Drives construction → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "enables-industry", "type": "causal", "position": {"x": 300, "y": 240}, "data": {"label": "Enables industry → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "needs-roads", "type": "causal", "position": {"x": 400, "y": 300}, "data": {"label": "Needs roads/airports → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "embodied-use", "type": "causal", "position": {"x": 200, "y": 540}, "data": {"label": "Embodied + use", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "energy-demand", "type": "causal", "position": {"x": 700, "y": 360}, "data": {"label": "Energy demand → ↑ CO₂", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "generation", "type": "causal", "position": {"x": 900, "y": 420}, "data": {"label": "Generation", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "power-generation", "type": "causal", "position": {"x": 900, "y": 540}, "data": {"label": "Power generation (SO₂, NOx)", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "direct-indirect", "type": "causal", "position": {"x": 50, "y": 340}, "data": {"label": "Direct + indirect", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "fuel-travel", "type": "causal", "position": {"x": 50, "y": 420}, "data": {"label": "Fuel & travel", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "co2", "type": "causal", "position": {"x": 300, "y": 650}, "data": {"label": "CO₂ Emissions", "value": 0, "enabled": true, "type": "output"}},
+    {"id": "contributes", "type": "causal", "position": {"x": 300, "y": 720}, "data": {"label": "Contributes to", "value": 0, "enabled": true, "type": "intermediate"}},
+    {"id": "aqi", "type": "causal", "position": {"x": 300, "y": 800}, "data": {"label": "Air Quality Index (AQI)", "value": 0, "enabled": true, "type": "output"}},
+    {"id": "poor-aqi", "type": "causal", "position": {"x": 700, "y": 650}, "data": {"label": "Poor AQI → Urban health/stability", "value": 0, "enabled": true, "type": "intermediate"}}
+  ],
+  "edges": [
+    {"id": "i-mg", "source": "industries", "target": "moves-goods", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "mg-t", "source": "moves-goods", "target": "transport", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "i-up", "source": "industries", "target": "uses-power", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "i-pi", "source": "industries", "target": "powers-industry", "data": {"weight": 0.4}, "type": "smoothstep"},
+    {"id": "pi-ip", "source": "powers-industry", "target": "industrial-pollutants", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "up-ip", "source": "uses-power", "target": "industrial-pollutants", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "pi-e", "source": "powers-industry", "target": "energy", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "t-ft", "source": "transport", "target": "fuels-transport", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "t-fr", "source": "transport", "target": "fuel-refining", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "ft-e", "source": "fuels-transport", "target": "energy", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "fr-e", "source": "fuel-refining", "target": "energy", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "t-us", "source": "transport", "target": "urban-sprawl", "data": {"weight": 0.4}, "type": "smoothstep"},
+    {"id": "t-ve", "source": "transport", "target": "vehicle-emissions", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "t-dc", "source": "transport", "target": "drives-construction", "data": {"weight": 0.4}, "type": "smoothstep"},
+    {"id": "t-ei", "source": "transport", "target": "enables-industry", "data": {"weight": 0.4}, "type": "smoothstep"},
+    {"id": "t-nr", "source": "transport", "target": "needs-roads", "data": {"weight": 0.4}, "type": "smoothstep"},
+    {"id": "dc-inf", "source": "drives-construction", "target": "infrastructure", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "ei-inf", "source": "enables-industry", "target": "infrastructure", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "nr-inf", "source": "needs-roads", "target": "infrastructure", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "us-inf", "source": "urban-sprawl", "target": "infrastructure", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "inf-eu", "source": "infrastructure", "target": "embodied-use", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "us-ed", "source": "urban-sprawl", "target": "energy-demand", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "ed-g", "source": "energy-demand", "target": "generation", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "e-g", "source": "energy", "target": "generation", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "g-pg", "source": "generation", "target": "power-generation", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "up-di", "source": "uses-power", "target": "direct-indirect", "data": {"weight": 0.5}, "type": "smoothstep"},
+    {"id": "ip-di", "source": "industrial-pollutants", "target": "direct-indirect", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "ve-ft", "source": "vehicle-emissions", "target": "fuel-travel", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "eu-co2", "source": "embodied-use", "target": "co2", "data": {"weight": 0.6}, "type": "smoothstep"},
+    {"id": "pg-co2", "source": "power-generation", "target": "co2", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "ve-co2", "source": "vehicle-emissions", "target": "co2", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "ip-co2", "source": "industrial-pollutants", "target": "co2", "data": {"weight": 0.8}, "type": "smoothstep"},
+    {"id": "di-co2", "source": "direct-indirect", "target": "co2", "data": {"weight": 0.8}, "type": "smoothstep"},
+    {"id": "ft-co2", "source": "fuel-travel", "target": "co2", "data": {"weight": 0.7}, "type": "smoothstep"},
+    {"id": "co2-c", "source": "co2", "target": "contributes", "data": {"weight": 0.8}, "type": "smoothstep"},
+    {"id": "c-aqi", "source": "contributes", "target": "aqi", "data": {"weight": 0.8}, "type": "smoothstep"},
+    {"id": "aqi-pa", "source": "aqi", "target": "poor-aqi", "data": {"weight": 0.8}, "type": "smoothstep"}
+  ]
+}

gunicorn_conf.py

@@ -0,0 +1,18 @@
+import os
+
+# Render.com gives 512MB RAM for the free tier.
+# We must limit workers to avoid OOM kills.
+# 1 worker + multiple threads is best for memory-constrained Python apps with I/O (like DB/LLM calls).
+
+port = os.environ.get("PORT", "10000")
+bind = f"0.0.0.0:{port}"
+
+workers = 1
+threads = 4
+timeout = 120  # LLM calls can be slow
+keepalive = 5
+
+# Logging
+accesslog = "-"
+errorlog = "-"
+loglevel = "info"

requirements.txt

@@ -33,3 +33,5 @@ xgboost
 folium
 selenium
 webdriver-manager
+shapely
+requests

train_aqi_model.py

@@ -0,0 +1,22 @@
+import os
+import joblib
+from aqi_history import AQIHistory
+
+MODEL_PATH = 'aqi_model.pkl'
+
+def train_and_save():
+    print("Initializing AQI History...")
+    # This triggers training in __init__
+    handler = AQIHistory()
+    
+    if handler.model:
+        print("Model trained successfully.")
+        print("Saving model to disk...")
+        joblib.dump(handler.model, MODEL_PATH)
+        joblib.dump(handler.metrics, 'aqi_metrics.pkl')
+        print(f"Model saved to {MODEL_PATH}")
+    else:
+        print("Failed to train model.")
+
+if __name__ == "__main__":
+    train_and_save()