Create model/anomaly_detector.py

model/anomaly_detector.py

@@ -0,0 +1,41 @@
+import numpy as np
+from sklearn.ensemble import IsolationForest
+from utils.preprocessing import extract_features
+
+# A placeholder Isolation Forest model
+# In production, load a pre-trained model using joblib or pickle
+model = IsolationForest(contamination=0.2, random_state=42)
+
+# Example training data (you should replace with real preprocessed dataset)
+example_training_data = np.array([
+    [5, 0, 30, 0],
+    [10, 1, 50, 1],
+    [2, 4, 10, 3],
+    [8, 0, 45, 0],
+    [3, 5, 12, 2]
+])
+model.fit(example_training_data)
+
+def detect_anomaly(agent_data):
+    features = extract_features(agent_data)
+    prediction = model.predict([features])[0]  # -1 for anomaly
+    flagged = prediction == -1
+
+    reason = ""
+    if flagged:
+        if agent_data["missed_visits"] >= 5:
+            reason = "Missed appointments spike"
+        elif agent_data["weekly_calls"] < 4:
+            reason = "Low call frequency"
+        elif agent_data["lead_drop_rate"] > 0.6:
+            reason = "High lead drop rate"
+        else:
+            reason = "Unusual activity pattern"
+
+    return {
+        "agent_id": agent_data.get("agent_id", "unknown"),
+        "weekly_calls": agent_data["weekly_calls"],
+        "missed_visits": agent_data["missed_visits"],
+        "flagged": flagged,
+        "reason": reason if flagged else "Normal behavior"
+    }