Estimating public demand following disasters through Bayesian-based information integration

Abstract

Public demand estimation is essential to effective relief resource distribution following disasters. However, previous studies are incapable of deriving a reliable estimation mainly due to the complexity, dynamicity, and nonlinearity of public demand. This research proposes an innovative data-driven approach to estimate public demand by leveraging sample information, such as social media and surveys. Twitter-based demand percentage (TDP) is designed as the predictor of actual demand percentage, while survey-based demand percentage (SDP) is developed as the ground truth of actual demand percentage. Sampling bias of social media users is removed through a systematic process that comprises the prediction of social media user races/ethnicities and the aggregation of demand percentages. Sampling uncertainty of TDP and SDP is modeled through a Bayesian-based approach that integrates prior knowledge as well as new observations from social media and surveys. The relationship between TDP and SDP is learned through a polynomial model, which facilitates the estimation of future actual demand percentage. To illustrate the feasibility and applicability of the proposed approach, public demand for COVID-19 vaccines in the US is estimated. Results demonstrate that the TDP is a strong predictor of actual demand percentage. This research novelly takes the advantages of sample information—the near-real-time nature of social media and the high reliability of surveys—to achieve a reliable and rapid estimation of public demand following disasters.