Total distance approximations for routing solutions

In order to make strategic, tactical and operational decisions, carriers and logistic companies need to evaluate scenarios with high levels of accuracy by solving a large number of routing problems. This might require relatively high computational efforts and time. In this paper, we present regression-based estimation models that provide fast predictions for the travel distance in the traveling salesman problem (TSP), the capacitated vehicle routing problem with Time Windows (CVRP-TW), and the multi-region multi-depot pickup and delivery problem (MR-MDPDP). The use of general characteristics such as distances, time windows, capacities and demands, allows us to extend the models and adjust them to different problems and also to different solution methods. The resulting regression models in most cases achieve good approximations of total travel distances except in cases where strong random noise is present, and outperform previous models.