2026-08-03 ノースカロライナ州立大学（NC State）
＜関連情報＞
- https://news.ncsu.edu/2026/08/more-efficient-optimization-modeling-for-allocation-problems/
- https://www.sciencedirect.com/science/article/pii/S2667259626000196
- https://www.sciencedirect.com/science/article/pii/S1093968726000939
- https://pubsonline.informs.org/doi/10.1287/trsc.2022.1134
機械学習を用いた最大被覆位置割り当て問題のための列生成 Machine learning-guided column generation for a maximal covering location–allocation problem
Kuangying Li, Hiruni Niwunhella, Leila Hajibabai, Ali Hajbabaie
Sustainability Analytics and Modeling Available online: 24 July 2026
DOI:https://doi.org/10.1016/j.samod.2026.100069
Abstract
This study presents a maximal covering location–allocation problem for vaccine distribution, formulated as a mixed-integer linear model. The model aims to minimize total distribution cost while maximizing vaccine allocation to population blocks. A modified Voronoi diagram technique embedded in a Lagrangian relaxation framework is employed to solve the problem. Leveraging real-world data from the Centers for Disease Control and Prevention (CDC) and health departments in Pennsylvania, empirical case studies validate the model’s effectiveness. Numerical experiments indicate that the proposed model outperforms traditional benchmark methods, including the column generation (CG) technique. However, both methods experience computational challenges due to large datasets and complex decision variables. To overcome such challenges, this study proposes a machine learning-enhanced column generation (ML-CG) framework. This hybrid strategy integrates offline training with online prediction to improve computational efficiency, reducing runtime while preserving solution quality with an acceptable optimality gap. The machine learning component is used to predict promising binary decision patterns for one sub-problem within the decomposition-based solution process. These prediction results are then incorporated into the CG procedure to support reduced-cost evaluation and improve the efficiency of column updates, rather than directly generating columns or replacing the exact optimization method. The numerical results show that the ML-CG framework accelerates runtime by 79.1% compared to the original CG approach while maintaining high-quality solutions. The enhanced performance resulting from integrating predictive analytics within the optimization framework improves the ability to efficiently address large-scale location–allocation problems with complex integer decision structures.
公平な資源配分のための緩和法に基づくボロノイ図アプローチ A relaxation‐based Voronoi diagram approach for equitable resource distribution
Kuangying Li, Asya Atik, Dayang Zheng, Leila Hajibabai, Ali Hajbabaie
Computer-Aided Civil and Infrastructure Engineering Available online: 15 March 2026
DOI:https://doi.org/10.1111/mice.13339
Abstract
This paper introduces a methodology designed to reduce cost, improve demand coverage, and ensure equitable vaccine distribution during the initial stages of the vaccination campaign when demand significantly exceeds supply. We formulate an enhanced maximum covering problem as a mixed integer linear program, aiming to minimize the total vaccine distribution cost while maximizing the allocation of vaccines to population blocks under equity constraints. Block‐level census data are employed to define demand locations, identifying gender, age, and racial groups within each block using population data. A Lagrangian relaxation technique integrated with a modified Voronoi diagram is proposed to solve the location–allocation problem efficiently. Empirical case studies in Pennsylvania, using real‐world data from the Centers for Disease Control and Prevention and health department websites, were conducted for the first 4 months of the COVID‐19 vaccination campaign. Preliminary results show that the proposed solution algorithm effectively solves the problem, achieving a 5.92% reduction in total transportation cost and a 28.15% increase in demand coverage. Moreover, our model can reduce the deviation from equity to 0.07 (∼50% improvement).
ワクチン出荷量と廃棄量に関するCOVID-19データを用いたモデリングと意思決定 Using COVID-19 Data on Vaccine Shipments and Wastage to Inform Modeling and Decision-Making
Leila Hajibabai ,Ali Hajbabaie ,Julie Swann ,Dan Vergano
Transportation Science Published: 1 Apr 2022
DOI:https://doi.org/10.1287/trsc.2022.1134
Abstract
Since the start of the COVID-19 pandemic, disruptions have been experienced in many supply chains, particularly in personal protective equipment, testing kits, and even essential household goods. Effective vaccines to protect against COVID-19 were approved for emergency use in the United States in late 2020, which led to one of the most extensive vaccination campaigns in history. We continuously collect data on vaccine allocation, shipment and distribution, administration, and inventory in the United States, covering the entire vaccination campaign. In this article, we describe some data sets that we collaborated to obtain. We are publishing the data and making them freely available to researchers, media organizations, and other stakeholders so that others may use the data to develop insights about the distribution and wastage of vaccines during the current pandemic or to provide an informed future pandemic response. This article gives an overview of vaccine distribution logistics in the United States, describes the data we obtain, outlines how they may be accessed and used by others, and describes some high-level analyses demonstrating some aspects of the data (for data collected during January 1, 2021–March 31, 2021). This article also provides directions for future research using the collected data. Our goal is two-fold: (i) We would like the data to be used in many creative ways to inform the current and future pandemic response. (ii) We also want to inspire other researchers to make their data publicly available in a timely manner.
