Optimization of Fresh Product Deliveries

Abstract

This final year project develops a dual-component logistics system for fresh food distribution, addressing demand forecasting and vehicle routing as separate problems. Conducted with Celia, it tackles operational challenges through datadriven solutions. The methodology includes demand forecasting with historical data, comparing traditional methods to Machine Learning (XGBoost), and vehicle routing optimization via mathematical modeling solved with IBM ILOG CPLEX. XGBoost yielded substantial improvements, while the VRP model achieved cost reductions and fleet efficiency. A web application integrates both components, providing strategic planning tools and practical optimization capabilities