FoodRL: A Reinforcement Learning Ensembling Framework For In-Kind Food Donation Forecasting

2025-11-06

AI Summary

Food banks face significant challenges—including high volatility in physical donations, seasonal fluctuations, and concept drift induced by natural disasters—undermining the robustness of conventional forecasting models. To address this, we propose a reinforcement learning–based meta-learning framework that dynamically clusters forecasting models and jointly incorporates contextual features and recent performance feedback to enable adaptive ensemble construction and optimal weight allocation. This approach effectively mitigates concept drift and substantially enhances prediction stability and generalization under unforeseen disruptions. Experiments on real-world operational data from two U.S. food banks demonstrate that our method achieves an average 12.3% improvement in forecasting accuracy over state-of-the-art baselines. Annually, it supports the precise reallocation of approximately 1.7 million meals, delivering a scalable, highly robust intelligent forecasting solution for humanitarian supply chains.

Technology Category

• Search and Optimization: Metareasoning and Metaheuristics
• Machine Learning: Ensemble Methods
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization

Abstract

Food banks are crucial for alleviating food insecurity, but their effectiveness hinges on accurately forecasting highly volatile in-kind donations to ensure equitable and efficient resource distribution. Traditional forecasting models often fail to maintain consistent accuracy due to unpredictable fluctuations and concept drift driven by seasonal variations and natural disasters such as hurricanes in the Southeastern U.S. and wildfires in the West Coast. To address these challenges, we propose FoodRL, a novel reinforcement learning (RL) based metalearning framework that clusters and dynamically weights diverse forecasting models based on recent performance and contextual information. Evaluated on multi-year data from two structurally distinct U.S. food banks-one large regional West Coast food bank affected by wildfires and another state-level East Coast food bank consistently impacted by hurricanes, FoodRL consistently outperforms baseline methods, particularly during periods of disruption or decline. By delivering more reliable and adaptive forecasts, FoodRL can facilitate the redistribution of food equivalent to 1.7 million additional meals annually, demonstrating its significant potential for social impact as well as adaptive ensemble learning for humanitarian supply chains.

Problem

• Forecasting volatile in-kind food donations for food banks
• Addressing unpredictable fluctuations and concept drift in donations
• Improving equitable resource distribution during disasters and disruptions

Innovation

• Reinforcement learning metalearning framework for forecasting
• Clusters and weights models dynamically by performance
• Adaptive ensemble learning for humanitarian supply chains