Using Machine Learning for Predictive Freight Demand and Route Optimization in Road and Rail Logistics

Abstract

About half of all line haul freight capacity in Europe is on trucks, which generate congestion, high infrastructure costs, and environmental degradation. That does not detract from the fact that transport is indispensable to economic progress, which includes the growth in welfare. The technology of machine learning is rapidly maturing. This builds on the broad availability of relevant data in the physical sphere of road and rail freight transport. The state of the art in this technology and its application in supporting better-informed decision-making in transport, with particular reference to truck and rail freight, is reviewed. The nature of prediction, the importance, and the complexities of various paradigms of pattern recognition, clustering of dependent observations, and path optimization are explained. It is discussed how these can drive growth in capacity by rendering transport more efficient and hence more cost-effective, without the need to compromise on environmental and social cohesion goals. All this requires a professional mindset, the establishment of skill sets, and the solution to ethical issues.