Publication Date
Authored by
Kaitlyn Rakestraw
Advisor(s): Inma Borrella
Topic(s) Covered:
  • Demand Planning

Food insecurity is a problem that affects people in every county within the United States. To combat this issue, many organizations across the country provide charitable food assistance to their communities. The demand for these services is variable, and many of these organizations do not have a consistent method of predicting future demand. This research explores the demand for one food bank, the Mid-Ohio Food Collective (MOFC), and analyzes how this demand differs based on the socioeconomic vulnerability of an area. Additionally, a variety of forecasting models are tested to determine which can best predict demand for the MOFC’s services, and the implications of improved forecast accuracy are investigated. This study suggests a framework for categorizing United States counties into two distinct groups, namely "more vulnerable" and "less vulnerable," utilizing socioeconomic factors. When applied to the case study of the MOFC, this classification allowed for identification of differences in demand patterns between the two clusters. Through an evaluation of the RMSE, MAE, and MAPE of nine time series forecasting models, it was found that a naive forecasting model performs well in forecasting demand for the more vulnerable counties in the MOFC’s service area. However, it was found that switching from a naive forecasting model to an exponential smoothing model with level and trend components can significantly improve demand forecasting accuracy for the less vulnerable counties. By switching to an exponential smoothing model with level and trend components for the less vulnerable cluster, the MOFC can improve their forecasting accuracy from a 9.9% MAPE using the naïve model to a 4.3%. The factors utilized in this study are relevant and applicable to all counties in the United States. As a result, the insights gained from this research can be effectively employed by food banks throughout the country, enabling them to improve the accuracy of their demand forecasts.