Estimating True Demand at Hunger Relief Organizations with Predictive Modeling

ABSTRACT

Significant uncertainty exists both on the supply and demand side of food bank operations, but the demand side has received significantly less attention in the literature. In this study, we used data from a food bank branch to characterize the behavior of their partner agencies. Of the three predictive models considered, random forest was applied to estimate the number of persons served by the partner agencies, which is then converted to food demand. Our bottom-up approach can be replicated for all branches in a food bank’s network to obtain their true demand.

KEYWORDS:

• Food demand
• predictive modeling
• food bank
• hunger relief
• random forest

Acknowledgments

We would like to express our sincere appreciation to the leadership of the Food Bank of Central and Eastern North Carolina (FBCENC) for their valuable input to this project. This research was supported by NSF grant (#1718672) titled PFI: BIC - Flexible, equitable, efficient, and effective distribution (FEEED) and NSF grant (#1735258) titled NRT: Improving strategies for hunger relief and food security using computational data science.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Supplementary material

Supplemental data for this article can be accessed on the publisher’s website

Additional information

Funding

This work was supported by the NSF [1718672].