Analysis of vehicle routing in a distribution network using machine learning techniques

This study presents the application of vehicle routing problem (VRP) in real-time. Vehicle routing minimises transportation costs by finding the optimal sequence for delivery in which all the customers are served. Today, implementing the VRP is still challenging due to its difficulty in terms of data collection, analysis, and optimisation. The main objective of this study is to analyse distinct types of distance metrics in solving the VRP and their relationship using machine learning (ML) techniques. Metaheuristic techniques are used to optimise vehicle routes and further combined with ML techniques to enhance solution quality. The prediction capability and solution quality using different techniques are analysed, and the computational result shows that distance metrics have a significant role in optimising vehicle routes and the adverse effect of using Euclidean distance in real-time. This study considers the most suitable real-time application area to implement VRP in a large customer market of essential commodities.