Multi-Criteria Decision Making With Machine Learning for Vehicle Routing Problem
This chapter addresses vehicle routing problem with time windows (VRPTW), one of the most well-known combinatorial optimization problems with many real-world applications in the transportation sector. This chapter proposes a three-stage approach for VRPTW and presents an application of this approach to a real-life problem. The stages of the approach include clustering of customers, determining feasible routes and their criteria values for each cluster, and selecting the best routes for each cluster based on multi-criteria decision analysis. In the first stage of the proposed approach, a fuzzy c-means (FCM) clustering-based assignment algorithm is used. The second stage includes predicting travel times between nodes based on GPS data with support vector regression (SVR) and applying the proposed feasible route determination and criteria value calculation algorithm using these predictions and other inputs. In the last stage, routes are selected with the analytic hierarchy process (AHP) and the technique for order of preference by similarity to ideal solution (TOPSIS) for each cluster.