Initial commit

app/services/ml/behavior_analyzer.py

@@ -0,0 +1,55 @@
+"""
+Feature band encoders
+======================
+Discrete bucketers for assignment-context features (distance, time-of-day, load,
+order density). These are small pure helpers used by the Thompson-sampling
+strategy bandit (`strategy_bandit.py`) and the /riderassign bandit context.
+
+NOTE: The ID3 SUCCESS/RISK decision tree that used to live here has been retired —
+it only ever produced response metadata and never affected any assignment. Only
+the generic feature-band encoders remain.
+"""
+
+from datetime import datetime
+
+
+def distance_band(km: float) -> str:
+    """Total route distance -> discrete band."""
+    if km <= 5.0:   return "SHORT"
+    if km <= 15.0:  return "MID"
+    if km <= 30.0:  return "LONG"
+    return "VERY_LONG"
+
+
+def time_band(ts_str: str) -> str:
+    """ISO timestamp -> time-of-day band."""
+    try:
+        hour = datetime.fromisoformat(ts_str).hour
+        if 6  <= hour < 10: return "MORNING_RUSH"
+        if 10 <= hour < 12: return "LATE_MORNING"
+        if 12 <= hour < 14: return "LUNCH_RUSH"
+        if 14 <= hour < 17: return "AFTERNOON"
+        if 17 <= hour < 20: return "EVENING_RUSH"
+        if 20 <= hour < 23: return "NIGHT"
+        return "LATE_NIGHT"
+    except Exception:
+        return "UNKNOWN"
+
+
+def load_band(avg_load: float) -> str:
+    """Average orders-per-rider -> load band."""
+    if avg_load <= 2.0:  return "LIGHT"
+    if avg_load <= 5.0:  return "MODERATE"
+    if avg_load <= 8.0:  return "HEAVY"
+    return "OVERLOADED"
+
+
+def order_density_band(num_orders: int, num_riders: int) -> str:
+    """Orders per available rider -> density band."""
+    if num_riders == 0:
+        return "NO_RIDERS"
+    ratio = num_orders / num_riders
+    if ratio <= 2.0:  return "SPARSE"
+    if ratio <= 5.0:  return "NORMAL"
+    if ratio <= 9.0:  return "DENSE"
+    return "OVERLOADED"