Express courier. Direct.
Also overnight.
x25pronto Switzerland takes goods direct to the destination. Small loads go by car. Larger loads go by van or Van XL. Direct, express and overnight trips run Monday to Sunday when a slot is free. Free times are shown in the booking calendar. This keeps the plan clear.
Clustering methods for urban collection routes
How to cut tour counts for urban pickup routes? We compare K‑Means, DBSCAN and geographic heuristics and show how planners in Zurich, Basel and Geneva can improve routes.
Quick answer
Short answer: Use a hybrid approach. K‑Means is fast for evenly spread pickup points. DBSCAN finds dense clusters and outliers. Geographic clustering (grid or Voronoi, street‑network based) respects barriers and yields more practical tour areas. A hybrid workflow typically reduces tour counts and drive time best.
Why cluster pickup points?
Clustering groups nearby pickups. That lowers stops per tour and cuts driving time. Dispatchers can define which orders to combine. This matters in dense Swiss cities like Zurich, Basel and Geneva where traffic and rules affect tours.
K‑Means: fast and scalable
K‑Means divides points into a set number k of groups. It is computationally cheap and works well when pickups are evenly distributed and you have a target number of tours. Drawback: it ignores geographical barriers (rivers, rail lines) and is sensitive to outliers. Use K‑Means for an initial macro split and refine with routing.
DBSCAN: density‑based clusters and outliers
DBSCAN groups points by density and flags isolated pickups as noise. Advantage: it yields natural hotspot clusters in dense inner‑city areas. Drawback: parameters (epsilon, MinPts) need tuning for each city and time window. DBSCAN is useful to separate hotspots into their own collection tours.
Geographic heuristics: grids, Voronoi and street networks
Geographic methods use spatial structure: grid cells or Voronoi polygons create simple service areas. More accurate is clustering along the street network using OSM or local maps. These approaches respect natural separations and produce more driver‑friendly tour boundaries.
Practical hybrid workflow
Recommended steps: 1) detect hotspots with DBSCAN; 2) produce a coarse split with K‑Means to match vehicle count; 3) refine cluster edges with a street‑network overlay or grid; 4) run route optimisation (TSP heuristics, respect time windows). This balances statistical robustness and operational reality.
Metrics to measure impact
Key KPIs: average stops per tour, driving time per tour, unplanned detours, and vehicle utilisation. Run A/B tests on historic orders and pilot districts such as Zurich districts, Basel city centre or Geneva suburbs.
Frequently asked questions
How many clusters should I pick?
It depends on fleet size, vehicle capacity and time windows. Start with the number of tours you want and adapt based on capacity constraints.
How to choose DBSCAN parameters?
Select epsilon from typical neighbour distance (e.g. 200–500 m in inner‑city areas). Set MinPts based on expected density, e.g. 3–6. Validate with silhouette scores or direct KPI comparisons.
When are geographic heuristics preferable?
When streets, waterways or operational rules create natural separations, or when point density varies strongly. They produce more practical service areas for drivers.
How do I test methods without operational risk?
Simulate on historical orders and run pilot weeks in selected neighbourhoods. Compare your KPIs before and after to measure improvements.