A Sustainable Performance Analysis Of Logistic And Route Planning Based On Machine Learning Algorithms

Abstract

Urban logistics is central to 21st-century supply chains because it promotes efficiency and care for the environment. The presented work provides a brief literature review on the subject and presents an integrated data approach based on the methodology of clustering and evolutionary algorithms to contribute to the enhancement of route planning and logistics location in the metropolis. This research uses geolocation data from 8,079 customer delivery points in Salvador, Brazil, the study aims to make a comparison of K-Means and Mean-Shift clustering techniques in defining service zones performance. Within clusters, intra-cluster routing is improved by shortest path algorithm that is Dijkstra and the GA. As indicated by the survey, integrating mean shift clustering and genetic algorithm provides the best solution for delivery, which reduces travel distance by 2-3 times (operational costs and CO₂ emissions by more than 40%). Similarly, the conclusion section reveals the potential of integrating different combined learning techniques to scale up the last-mile delivery solutions into more flexible and sustainable models.