A Genetic Algorithm Data-Driven Approach for Middle-Mile Delivery Optimization

This paper proposes an integrated approach using the Nearest Neighbor Heuristic (NNH), historical delivery data analysis, and Multi-Criteria Decision Making (MCDM) with Analytic Hierarchy Process (AHP) and Technique for Order of Preference by Similarity to Ideal Solution (TOPSIS) within a Genetic Algorithm (GA) to optimize routes for middle-mile delivery (MMD) in the Closed-Open Mixed Vehicle Routing Problem (COMVRP). By leveraging delivery history and managerial input, this method facilitates optimal route generation, aiding logistics planning and decision-making. Focused on middle-mile distribution in the postal industry, the study addresses challenges of meeting increasing customer demands for faster delivery despite lower volumes. The integration of GA and MCDM improves performance, reducing route mileage, costs, and the number of required vehicles. Applied to Pos Indonesia's historical delivery dataset, the AHP-TOPSIS-Monte Carlo Simulation method effectively prioritizes MMD nodes, allowing decision-makers to define optimal route configurations, leading to a 36.38% increase in precision percentage node priorities and a reduction of 133 kilometers in route mileage. This study contributes to MMD optimization and supports decision-making through MCDM, COMVRP, and simulation integration.