Creation of Optimal Service Zones for the Delivery of Express Packages

Understanding postal delivery areas in the Republic of Korea using multiple unsupervised learning approaches

“…In practice, however, most of the service zones were manually defined by referring to the subdivisions designed by other operators for similar services, such as ZIP codes or city districts. These caused highly unbalanced workloads due to the geographical variants of the zones [21]. Wong [22] introduced some practical approaches for delivery zone partitioning and addressed again that parcel pickup and delivering are vital components of national and international transportation.…”

“…Schneider and others [25] focused on the influence of time window constraints on delivery routing and addressed that considering geographical aspects in districts is significant for generating high‐quality territories, whereas explicitly incorporating time window characteristics and historical demand data does not lead to a perceptible improvement of the solution quality. Parariani and others [21] proposed a decision support system that partitioned package delivery zones by exploiting an advanced multi‐attribute clustering algorithm, a variant of the k ‐means algorithm to match between customers and clusters. Barzegar and others [26] exploited expert knowledge of seasoned individuals, such as taxi drivers to find low‐traffic routes in big cities by constructing a knowledge base in a form of ontology.…”