Update app.py

app.py

@@ -7,7 +7,7 @@ from prophet import Prophet
 import networkx as nx
 from eventlet import event
 
-# Generate synthetic data
+# Generate synthetic data with more complexity
 def generate_data():
     schools = [f"School_{i}" for i in range(1, 11)]  # 10 Schools
     time_range = [datetime.now() - timedelta(hours=i) for i in range(24)]  # Past 24 hours
@@ -15,7 +15,8 @@ def generate_data():
     data = []
     for school in schools:
         for time in time_range:
-            bandwidth_usage = np.random.randint(1, 100)  # Random bandwidth usage in Mbps
+            bandwidth_usage = np.random.normal(loc=50, scale=20)  # Normal distribution for more variation
+            bandwidth_usage = max(1, min(150, bandwidth_usage))  # Keep within a realistic range
             data.append([school, time, bandwidth_usage])
     
     df = pd.DataFrame(data, columns=["School", "Timestamp", "Bandwidth_Usage"])
@@ -26,8 +27,11 @@ def generate_data():
 df = generate_data()
 
 def detect_anomalies(df):
-    df['Z_Score'] = (df['Bandwidth_Usage'] - df['Bandwidth_Usage'].mean()) / df['Bandwidth_Usage'].std()
-    anomalies = df[df['Z_Score'].abs() > 2]  # Mark as anomaly if Z-score > 2
+    # Using IQR for anomaly detection
+    Q1 = df['Bandwidth_Usage'].quantile(0.25)
+    Q3 = df['Bandwidth_Usage'].quantile(0.75)
+    IQR = Q3 - Q1
+    anomalies = df[(df['Bandwidth_Usage'] < (Q1 - 1.5 * IQR)) | (df['Bandwidth_Usage'] > (Q3 + 1.5 * IQR))]
     return anomalies
 
 def train_prophet(df):