Proposal
Over 44 million Americans currently suffer from food insecurity, of whom 13 million are children. Food insecurity has been shown to cause a wide range of both physical and developmental issues. Across the United States, thousands of food banks and pantries serve as vital sources of food and other forms of aid for food-insecure families. By optimizing food bank locations, food banks and their resources would become more accessible to families who desperately require it. The aim of this paper is to build a machine learning framework that is able to optimize food bank locations and to consider factors such as median income. We utilized the K-means clustering algorithm for this purpose due to its high processing speed and ability to factor in large amounts of data, as well as its unsupervised nature that doesn’t require training time or labeled training data. Our proposed method applies K-means over a range of houses sourced from California and Indiana U.S. Census and geospatial data, with a weighted K-means algorithm applied when income data is available. We generated food bank locations that aimed to prioritize lower income households and compared these locations against real food banks affiliated with Feeding America. Our results show that not only is K-means extremely fast, but our food bank locations were on average better than those of existing ones, saving distance between food banks and houses in both California and Indiana.
Recommended Citation
Ruan, Gavin
(2024)
"Where to Build Food Banks: A Machine Learning Approach,"
The Journal of Purdue Undergraduate Research:
Vol. 14, Article 11.
DOI: https://doi.org/10.7771/2158-4052.1661
