A data-driven decision support system for service completion prediction in last mile logistics
The growing demand for last mile services (deliveries and pickups) often results in the work overload of couriers, who are unable to complete all their assigned services within their working day. Uncompleted services are a source of strong dissatisfaction by customers, particularly since they were p...
| Autores: | , , , |
|---|---|
| Tipo de recurso: | artículo |
| Estado: | Versión aceptada para publicación |
| Fecha de publicación: | 2023 |
| País: | España |
| Institución: | Universidad de Sevilla (US) |
| Repositorio: | idUS. Depósito de Investigación de la Universidad de Sevilla |
| OAI Identifier: | oai:idus.us.es:11441/182854 |
| Acceso en línea: | https://hdl.handle.net/11441/182854 https://doi.org/10.1016/j.tra.2023.103817 |
| Access Level: | acceso abierto |
| Palabra clave: | Last mileClusteringRoutingMachine learning Machine learning Routing Clustering |
| Sumario: | The growing demand for last mile services (deliveries and pickups) often results in the work overload of couriers, who are unable to complete all their assigned services within their working day. Uncompleted services are a source of strong dissatisfaction by customers, particularly since they were probably aware that their requested service was scheduled for the day. The possibility of predicting how many and which are going to be these uncompleted services becomes an effective decision-making tool that would allow carriers to increase their perceived service levels without increasing the number of couriers and vehicles. This issue is addressed through the combination of two models. Firstly, machine learning techniques are applied to estimate how many services will remain uncompleted on a given route. Secondly, the use of clustering techniques is proposed as the basis to predict the routes to be followed by couriers, thus identifying potentially uncompleted services as the last ones in each route. The posited methodology is illustrated with a case study comprising four regions in Spain, obtaining promising results in terms of the predictive capacity and the accuracy of the models. |
|---|