Transportation Models and Unsupervised Learning Methods to Minimize Food Waste in Schools of Medellin

This research creates alternative vehicle routing plans for schools and marketplaces using unsupervised transportation models and learning methods. This initiative seeks to optimize food distribution logistics and reduce waste, thereby improving access to fresh food and mitigating food insecurity. The methodology involves the application of transportation models and unsupervised learning techniques to develop efficient vehicle routing plans. This study utilizes the K-means algorithm to identify vulnerable and privileged areas in Medellín, focusing on schools and marketplaces. Data collection and preparation are crucial steps, ensuring that the information accurately reflects the realities of food distribution. The gravity method is also employed to optimize delivery routes by considering factors such as distance, traffic, and vehicle load capacity. The vulnerable areas and the establishment of collection points for food aid were identified. The K-means algorithm helped determine centroids for both distribution and collection centers, facilitating the planning of logistics for food redistribution. A detailed map was created to visualize these zones, which will serve as essential tools for planning relief activities. The study also highlights the significant amount of food wasted in Medellín's marketplaces, emphasizing the need for effective routing to minimize this waste. The project achieved its objectives by identifying strategic locations for distribution centers and optimizing delivery routes. However, it faced challenges such as the lack of real-time data on population and food distribution. Future recommendations include integrating advanced data collection technologies and implementing a route management system using machine learning algorithms to adapt to changing conditions.