Dynamic items delivery network: prediction and clustering

Items delivery companies generally use a model to minimize delivery costs. From a mathematical perspective, the model is an objective function that involves constraints. Meanwhile, from a practical point of view, these constraints include aspects that affect item delivery, for example, delivery zones, number of delivery vehicles, vehicle capacity, trip routes, etc. However, the models built so far have not paid attention to changes in road density. This aspect can result in a nonoptimal delivery model, which results in not a minimum delivery cost. For this reason, this paper discusses how to divide zones using the clustering method and predict changes in the shipping zone of a dynamic network using predictive distribution. So, the model can work optimally if the delivery zones and delivery strategies are suitable.