The Trans-Disciplinary Use of Advanced Analytics to Improve Granular Accuracy in Identifying Food-Insecure Population, their Socio-Economic Characteristics, and Predict Food Service Demand at the Neighborhood Level

Keywords: food deserts, urban food systems, food insecurity
Session Type: Virtual Paper Abstract
Day: Friday
Session Start / End Time: 2/25/2022 02:00 PM (Eastern Time (US & Canada)) - 2/25/2022 03:20 PM (Eastern Time (US & Canada))
Room: Virtual 56

Abstract

Food-insecure areas in the US metropolitan regions face substantial challenges in identifying precise locations of vulnerable population and to accurately estimate the demand for food related services. Currently available identification methods, such as the ones used by USDA, rely on broad strokes to delineate the entire ZIP codes or Census Tracts as ‘food deserts’. This method is less reliable in making precise demand estimates causing already resource-strapped service providers to adopt arbitrary estimation, while simultaneously ignoring other forms of non- and less-corporate food services. In absence of an accurate data-centric estimation method, substantial misalignment exists between the service providers’ perception of demand and the actual service needs of the people in food-insecure neighborhoods.

This paper, the effort a trans-disciplinary collaborative team of scholars, focuses on using granular geospatial data and advanced analytics to identify such neighborhoods, their socio-economic characteristics, and predict the demand for food service. Our pilot stage analysis of the Richmond, Virginia region (1) delineates food insecurity areas at census block level, and (2) estimates food service demand (e.g. pound per week) at the household level. This predictive modeling draws upon a much larger dataset than traditionally used in food desert mapping including: diverse forms of food retail (grocery stores, bodega, community farms); network data of roads and sidewalks; demographic and socio-economic data from the American Community Survey (ACS); tax parcel data; and public transit service data. The outcome of this exercise is expected to provide accurate estimation civil society and private sector actors make operational decisions.