Update solver.py

solver.py

@@ -113,7 +113,13 @@ def tw_aware_clusters(df: pd.DataFrame, depot: Tuple[float, float],
     ang = np.arctan2(dy, dx)
 
     distances = np.sqrt(dx**2 + dy**2)
-    tw_urgency = df["tw_end"].values / (distances + 1.0)
+    #tw_urgency = df["tw_end"].values / (distances + 1.0)
+    # Earlier deadlines (smaller tw_end) → higher urgency
+    # Shorter time windows are also treated as slightly more urgent
+    tw_window = df["tw_end"].values - df["tw_start"].values
+    tw_urgency = (1.0 / (df["tw_end"].values + 1.0)) * (1.0 + 1.0 / (tw_window + 1.0))
+    tw_urgency = tw_urgency / (distances + 1.0)
+
     order = np.lexsort((tw_urgency, ang))
 
     clusters = [[] for _ in range(n_vehicles)]
@@ -181,10 +187,20 @@ def build_route_by_insertion_tw(df: pd.DataFrame, idxs: List[int],
         return []
     route, remaining = [], set(idxs)
 
+    """
     def urgency_score(i):
         dist = euclid(depot, (df.loc[i, "x"], df.loc[i, "y"]))
         tw_e = float(df.loc[i, "tw_end"])
-        return tw_e / (dist + 1.0)
+        return tw_e / (dist + 1.0)"""
+
+    def urgency_score(i):
+    dist = euclid(depot, (df.loc[i, "x"], df.loc[i, "y"]))
+    tw_s = float(df.loc[i, "tw_start"])
+    tw_e = float(df.loc[i, "tw_end"])
+    tw_window = max(1.0, tw_e - tw_s)
+    # Earlier deadlines and tighter windows → higher urgency (lower numeric score)
+    return (1.0 / (tw_e + 1.0)) * (1.0 + 1.0 / (tw_window + 1.0)) / (dist + 1.0)
+
 
     first = min(remaining, key=urgency_score)
     route.append(first)