Update app.py

app.py

@@ -16,42 +16,33 @@ st.set_page_config(
 # 2. PROFESSIONAL STYLING (THEME OVERRIDE)
 st.markdown("""
     <style>
-        /* IMPORT FONTS */
         @import url('https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700&display=swap');
         
-        /* FORCE LIGHT THEME BACKGROUNDS & TEXT */
         .stApp {
-            background-color: #f8fafc; /* Light Blue-Grey */
-            color: #0f172a; /* Slate 900 */
+            background-color: #f8fafc;
+            color: #0f172a;
             font-family: 'Inter', sans-serif;
         }
 
-        /* METRIC CARDS - GLASSMORPHISM */
+        /* METRIC CARDS */
         div[data-testid="stMetric"] {
             background-color: #ffffff;
             border: 1px solid #e2e8f0;
             border-radius: 8px;
             padding: 15px;
             box-shadow: 0 4px 6px -1px rgba(0, 0, 0, 0.1);
-            transition: transform 0.2s;
         }
-        div[data-testid="stMetric"]:hover {
-            transform: translateY(-2px);
-            box-shadow: 0 10px 15px -3px rgba(0, 0, 0, 0.1);
-        }
-        
-        /* FORCE DARK TEXT FOR METRICS (Fixes White-on-White) */
         div[data-testid="stMetricValue"] {
             color: #0f172a !important; 
             font-weight: 700 !important;
         }
         div[data-testid="stMetricLabel"] {
-            color: #64748b !important; /* Slate 500 */
+            color: #64748b !important;
         }
 
-        /* DATAFRAME STYLING (Fixes White-on-White) */
+        /* DATAFRAME */
         div[data-testid="stDataFrame"] div[role="grid"] {
-            color: #334155 !important; /* Slate 700 */
+            color: #334155 !important;
             background-color: white !important;
         }
         div[data-testid="stDataFrame"] div[role="columnheader"] {
@@ -60,25 +51,20 @@ st.markdown("""
             background-color: #f1f5f9 !important;
         }
 
-        /* SIDEBAR STYLING */
+        /* SIDEBAR */
         [data-testid="stSidebar"] {
-            background-color: #1e293b; /* Slate 800 */
+            background-color: #1e293b;
         }
         [data-testid="stSidebar"] * {
-            color: #f8fafc !important; /* Light text for sidebar */
+            color: #f8fafc !important;
         }
         [data-testid="stSidebar"] .stSelectbox label, 
         [data-testid="stSidebar"] .stMultiSelect label {
             color: #94a3b8 !important;
         }
 
-        /* HEADERS */
-        h1, h2, h3 {
-            color: #0f172a !important;
-            font-weight: 700 !important;
-        }
+        h1, h2, h3 { color: #0f172a !important; font-weight: 700 !important; }
         
-        /* CUSTOM BADGES */
         .status-badge {
             display: inline-flex;
             align-items: center;
@@ -90,26 +76,24 @@ st.markdown("""
         .bg-red { background-color: #fee2e2; color: #991b1b; }
         .bg-green { background-color: #dcfce7; color: #166534; }
         
-        /* MAP CANVAS FIX */
-        .js-plotly-plot .plotly .main-svg {
-            background-color: rgba(0,0,0,0) !important;
-        }
+        .js-plotly-plot .plotly .main-svg { background-color: rgba(0,0,0,0) !important; }
     </style>
 """, unsafe_allow_html=True)
 
 # 3. SMART DATA LOADING (MAPPING)
 @st.cache_data
 def load_data():
-    # 1. Load or Generate Data
     try:
         df = pd.read_csv('analyzed_aadhaar_data.csv')
     except FileNotFoundError:
-        # Dummy Data Generator if file missing
         dates = pd.date_range(start="2025-01-01", periods=200)
+        # Using realistic district names that imply direction for testing semantic logic
+        districts = ['North District', 'South Region', 'East Zone', 'West End', 'Central Hub', 'Rural A', 'Urban B']
+        
         df = pd.DataFrame({
             'date': dates,
-            'state': np.random.choice(['Maharashtra', 'Uttar Pradesh', 'Bihar', 'Karnataka', 'Delhi', 'West Bengal', 'Tamil Nadu', 'Gujarat', 'Rajasthan', 'Kerala'], 200),
-            'district': np.random.choice(['North', 'South', 'East', 'West', 'Central', 'Rural A', 'Urban B'], 200),
+            'state': np.random.choice(['Maharashtra', 'Uttar Pradesh', 'Bihar', 'Karnataka', 'Delhi', 'West Bengal', 'Kerala', 'Assam', 'Rajasthan', 'Gujarat'], 200),
+            'district': np.random.choice(districts, 200),
             'pincode': np.random.randint(110001, 800000, 200),
             'RISK_SCORE': np.random.uniform(15, 99, 200),
             'total_activity': np.random.randint(50, 800, 200),
@@ -118,12 +102,10 @@ def load_data():
             'is_weekend': np.random.choice([0, 1], 200, p=[0.7, 0.3])
         })
 
-    # Standardize Date
     if 'date' in df.columns:
         df['date'] = pd.to_datetime(df['date'])
     
-    # SMART GEO-CLUSTERING LOGIC
-    # Comprehensive Center Points for Indian States & UTs
+    # --- 1. PRECISE GEOMETRIC CENTERS ---
     state_centers = {
         'Andaman and Nicobar Islands': (11.7401, 92.6586),
         'Andhra Pradesh': (15.9129, 79.7400),
@@ -137,7 +119,7 @@ def load_data():
         'Goa': (15.2993, 74.1240),
         'Gujarat': (22.2587, 71.1924),
         'Haryana': (29.0588, 76.0856),
-        'Himachal Pradesh': (31.9579, 77.1095),
+        'Himachal Pradesh': (31.9579, 77.1095), # Corrected
         'Jammu and Kashmir': (33.7782, 76.5762),
         'Jharkhand': (23.6102, 85.2799),
         'Karnataka': (15.3173, 75.7139),
@@ -163,70 +145,78 @@ def load_data():
         'West Bengal': (22.9868, 87.8550)
     }
 
+    # --- 2. ANISOTROPIC SPREADS (Shape of the State) ---
+    # format: (lat_spread, lon_spread) in degrees
+    # This prevents "Thin" states from spilling into the ocean/neighbors
+    state_spreads = {
+        'Kerala': (1.2, 0.25),      # Tall and Thin
+        'West Bengal': (1.4, 0.4),  # Tall and Thin
+        'Assam': (0.4, 1.8),        # Wide
+        'Maharashtra': (1.2, 2.0),  # Wide
+        'Uttar Pradesh': (1.0, 2.2),# Wide
+        'Bihar': (0.8, 1.5),        # Wide
+        'Delhi': (0.08, 0.1),       # Tiny
+        'Goa': (0.15, 0.15),        # Tiny
+        'Chandigarh': (0.03, 0.03), # City
+        'Gujarat': (1.0, 1.3),
+        'Rajasthan': (1.8, 1.8),
+        'Madhya Pradesh': (1.5, 2.0),
+        'Andaman and Nicobar Islands': (1.5, 0.2), # Archipelago (Tall)
+        'Himachal Pradesh': (0.5, 0.6)
+    }
+    
+    default_spread = (0.6, 0.6)
+
     def get_coords(row):
         state = row.get('state', 'Delhi')
         district = str(row.get('district', 'Unknown'))
         
-        # 1. Get State Base Coordinates (Use your updated list)
         base_lat, base_lon = state_centers.get(state, (20.5937, 78.9629))
+        lat_scale, lon_scale = state_spreads.get(state, default_spread)
         
-        # 2. DEFINE STATE RADIUS SCALER (In Degrees)
-        # Default is 0.5 (~55km) which is safer than 1.5
-        default_radius = 0.5 
+        # --- 3. SEMANTIC OFFSETTING ---
+        # If district name contains direction, bias the jitter
+        lat_bias, lon_bias = 0, 0
+        d_lower = district.lower()
         
-        # Tighter constraints for small States/UTs
-        radius_map = {
-            'Chandigarh': 0.04,
-            'Delhi': 0.15,
-            'Goa': 0.15,
-            'Puducherry': 0.1,
-            'Lakshadweep': 0.05,
-            'Daman and Diu': 0.05,
-            'Dadra and Nagar Haveli': 0.05,
-            'Kerala': 0.3,      # Narrow state
-            'Haryana': 0.4,
-            'Punjab': 0.4,
-            'Tripura': 0.3,
-            'Sikkim': 0.15,
-            'Andaman and Nicobar Islands': 1.0 # Long archipelago
-        }
+        # Bias factor (percent of scale)
+        bias_factor = 0.7 
         
-        # Get the specific radius for this state
-        radius = radius_map.get(state, default_radius)
+        if 'north' in d_lower: lat_bias += lat_scale * bias_factor
+        if 'south' in d_lower: lat_bias -= lat_scale * bias_factor
+        if 'east' in d_lower:  lon_bias += lon_scale * bias_factor
+        if 'west' in d_lower:  lon_bias -= lon_scale * bias_factor
         
-        # 3. DETERMINISTIC HASHING
+        # --- 4. DETERMINISTIC RANDOMNESS ---
         district_hash = hash(state + district)
         np.random.seed(district_hash % 2**32) 
         
-        # Offset using the specific radius
-        dist_lat_offset = np.random.uniform(-radius, radius)
-        dist_lon_offset = np.random.uniform(-radius, radius)
+        # Random component (reduced if bias is present to keep it focused)
+        random_factor = 0.5 if (lat_bias != 0 or lon_bias != 0) else 1.0
         
-        # 4. MICRO JITTER (Random noise for individual points)
-        np.random.seed(None) 
-        noise_lat = np.random.normal(0, 0.02 * radius) # Scale noise relative to state size
-        noise_lon = np.random.normal(0, 0.02 * radius)
+        dist_lat = np.random.uniform(-lat_scale * random_factor, lat_scale * random_factor)
+        dist_lon = np.random.uniform(-lon_scale * random_factor, lon_scale * random_factor)
+        
+        # --- 5. MICRO NOISE (Avoid overlapping dots) ---
+        np.random.seed(None)
+        noise = 0.03
         
         return pd.Series({
-            'lat': base_lat + dist_lat_offset + noise_lat,
-            'lon': base_lon + dist_lon_offset + noise_lon
+            'lat': base_lat + lat_bias + dist_lat + np.random.normal(0, noise),
+            'lon': base_lon + lon_bias + dist_lon + np.random.normal(0, noise)
         })
     
-    # Apply coordinates
     coords = df.apply(get_coords, axis=1)
     df['lat'] = coords['lat']
     df['lon'] = coords['lon']
     
-    # Risk Categories
     df['risk_category'] = pd.cut(
         df['RISK_SCORE'],
         bins=[-1, 50, 75, 85, 100],
         labels=['Low', 'Medium', 'High', 'Critical']
     )
-    
     return df
 
-# Load Data
 df = load_data()
 
 # 4. SIDEBAR & FILTERS
@@ -234,11 +224,9 @@ with st.sidebar:
     st.markdown("### S.T.A.R.K AI Control")
     st.markdown("---")
     
-    # State Filter
     state_list = ['All'] + sorted(df['state'].unique().tolist())
     selected_state = st.selectbox("Select State", state_list)
     
-    # District Filter
     if selected_state != 'All':
         filtered_df = df[df['state'] == selected_state]
         district_list = ['All'] + sorted(filtered_df['district'].unique().tolist())
@@ -252,32 +240,24 @@ with st.sidebar:
         filtered_df = filtered_df[filtered_df['district'] == selected_district]
         
     st.markdown("---")
-    
-    # Risk Filter
     risk_filter = st.multiselect(
         "Risk Level",
         options=['Low', 'Medium', 'High', 'Critical'],
         default=['High', 'Critical']
     )
-    
     if risk_filter:
         filtered_df = filtered_df[filtered_df['risk_category'].isin(risk_filter)]
     
     st.markdown("---")
-    
-    # Links
-    st.markdown("**Resources**")
     st.link_button("Open Notebook in Colab", "https://colab.research.google.com/drive/1YAQ4nfxltvG_cts3fmGc_zi2JQc4oPOT?usp=sharing")
-
     st.markdown("---")
     st.info(f"**User:** UIDAI_Officer\n\n**Team:** UIDAI_4571")
 
-# 5. HEADER & KPI METRICS
+# 5. HEADER
 col1, col2 = st.columns([3, 1])
 with col1:
     st.title("Project S.T.A.R.K AI Dashboard")
     st.markdown("Context-Aware Fraud Detection System")
-
 with col2:
     st.markdown("""
     <div style="text-align: right; padding-top: 20px;">
@@ -288,7 +268,7 @@ with col2:
 
 st.markdown("---")
 
-# METRICS ROW
+# METRICS
 m1, m2, m3, m4 = st.columns(4)
 total_centers = len(filtered_df)
 high_risk = len(filtered_df[filtered_df['RISK_SCORE'] > 75])
@@ -300,29 +280,24 @@ m2.metric("High Risk Alerts", f"{high_risk}", delta="Action Required", delta_col
 m3.metric("Avg. Risk Score", f"{avg_risk:.1f}/100", border=True)
 m4.metric("Weekend Spikes", f"{weekend_alerts}", "Unauthorized", delta_color="off", border=True)
 
-st.markdown("##") # Spacer
+st.markdown("##")
 
-# 6. MAIN TABS
+# 6. TABS
 tab_map, tab_list, tab_charts = st.tabs(["Geographic Risk", "Priority List", "Pattern Analytics"])
 
-# TAB 1: GEOGRAPHIC RISK (MAP)
 with tab_map:
     col_map, col_details = st.columns([3, 1])
-    
     with col_map:
         if not filtered_df.empty:
-            # Using Open-Street-Map for better contrast and no-token requirement
             fig_map = px.scatter_mapbox(
                 filtered_df,
-                lat="lat",
-                lon="lon",
+                lat="lat", lon="lon",
                 color="RISK_SCORE",
                 size="total_activity",
-                # Traffic Light Colors: Green -> Yellow -> Red
                 color_continuous_scale=["#22c55e", "#eab308", "#ef4444"], 
                 size_max=20,
                 zoom=4.5 if selected_state != 'All' else 3.5,
-                center={"lat": 22.0, "lon": 80.0}, # Center of India
+                center={"lat": 22.0, "lon": 80.0},
                 hover_name="pincode",
                 hover_data={"district": True, "state": True, "RISK_SCORE": True, "lat": False, "lon": False},
                 mapbox_style="open-street-map", 
@@ -339,7 +314,6 @@ with tab_map:
         if not filtered_df.empty:
             top_districts = filtered_df.groupby('district')['RISK_SCORE'].mean().sort_values(ascending=False).head(5)
             for district, score in top_districts.items():
-                # Color code the side bar
                 color = "#ef4444" if score > 80 else "#f59e0b"
                 st.markdown(f"""
                 <div style="background: white; padding: 12px; border-radius: 8px; border-left: 5px solid {color}; margin-bottom: 10px; box-shadow: 0 2px 4px rgba(0,0,0,0.05);">
@@ -348,80 +322,41 @@ with tab_map:
                 </div>
                 """, unsafe_allow_html=True)
 
-# TAB 2: PRIORITY LIST (DATAFRAME)
 with tab_list:
     st.subheader("Target Investigation List")
     st.markdown("Filter: *Showing centers with Risk Score > 75*")
-    
     target_list = filtered_df[filtered_df['RISK_SCORE'] > 75].sort_values('RISK_SCORE', ascending=False)
-    
     st.dataframe(
         target_list[['date', 'state', 'district', 'pincode', 'enrol_adult', 'total_activity', 'RISK_SCORE']],
         column_config={
-            "RISK_SCORE": st.column_config.ProgressColumn(
-                "Risk Probability",
-                help="Probability of fraud based on context analysis",
-                format="%d%%",
-                min_value=0,
-                max_value=100,
-            ),
+            "RISK_SCORE": st.column_config.ProgressColumn("Risk Probability", format="%d%%", min_value=0, max_value=100),
             "date": st.column_config.DateColumn("Date", format="DD MMM YYYY"),
             "total_activity": st.column_config.NumberColumn("Volume"),
             "enrol_adult": st.column_config.NumberColumn("Adult Enrols"),
         },
-        use_container_width=True,
-        hide_index=True,
-        height=400
+        use_container_width=True, hide_index=True, height=400
     )
-    
-    # Export Button
     csv = target_list.to_csv(index=False).encode('utf-8')
-    st.download_button(
-        "Download CSV",
-        data=csv,
-        file_name="uidai_stark_ai_priority_list.csv",
-        mime="text/csv",
-        type="primary"
-    )
+    st.download_button("Download CSV", data=csv, file_name="uidai_stark_priority_list.csv", mime="text/csv", type="primary")
 
-# --- TAB 3: CHARTS ---
 with tab_charts:
     c1, c2 = st.columns(2)
-    
     with c1:
-        st.subheader("Ghost ID Pattern (Ratio Deviation)")
-        # Scatter Plot
+        st.subheader("Ghost ID Pattern")
         fig_scatter = px.scatter(
-            filtered_df,
-            x="total_activity",
-            y="ratio_deviation",
+            filtered_df, x="total_activity", y="ratio_deviation",
             color="risk_category",
             color_discrete_map={'Critical': '#ef4444', 'High': '#f97316', 'Medium': '#eab308', 'Low': '#22c55e'},
             title="Deviation from District Baseline",
-            labels={"ratio_deviation": "Deviation Score", "total_activity": "Daily Transactions"},
-            hover_data=['pincode', 'district']
+            labels={"ratio_deviation": "Deviation Score", "total_activity": "Daily Transactions"}
         )
         fig_scatter.add_hline(y=0.2, line_dash="dash", line_color="red", annotation_text="Fraud Threshold")
         st.plotly_chart(fig_scatter, use_container_width=True)
-
     with c2:
         st.subheader("Risk Distribution")
-        # Histogram
-        fig_hist = px.histogram(
-            filtered_df, 
-            x="RISK_SCORE", 
-            nbins=20,
-            color_discrete_sequence=['#3b82f6'],
-            title="Frequency of Risk Scores"
-        )
+        fig_hist = px.histogram(filtered_df, x="RISK_SCORE", nbins=20, color_discrete_sequence=['#3b82f6'], title="Frequency of Risk Scores")
         fig_hist.update_layout(bargap=0.1)
         st.plotly_chart(fig_hist, use_container_width=True)
 
-# 7. FOOTER
 st.markdown("---")
-st.markdown("""
-<div style="text-align: center; font-size: 13px; color: #94a3b8;">
-    <b>Project S.T.A.R.K AI</b> | UIDAI Hackathon 2026 | Team UIDAI_4571<br>
-    <i>Confidential - For Official Use Only</i>
-</div>
-""", unsafe_allow_html=True)
+st.markdown("""<div style="text-align: center; font-size: 13px; color: #94a3b8;"><b>Project S.T.A.R.K AI</b> | UIDAI Hackathon 2026</div>""", unsafe_allow_html=True)