An Individual-Oriented Typology of Social Areas in the United States

Geodemographics, the practice of characterizing neighborhood socio-cultural landscapes, aids decision making in a variety of planning and administrative domains, including emergency management, transportation and energy use, and public health. This research establishes a new individual-oriented framework for geodemographic classification of neighborhoods in the United States.  Our framework is “individual-oriented” in that it leverages responses from the American Community Survey’s Public-Use Microdata Sample (PUMS) to create a topology of people, households, and neighborhoods. We achieve this first by identifying key population segments from the PUMS (termed “cohort profiles”), then identifying how the social landscape is differentiated by the co-location of these segments (termed “neighborhood profiles”).

Our approach advances methods for geodemographic classification in several key ways. First, individual-level information provides a greater flexibility for combining variables on people and households, enabling bespoke classifications for different modeling scenarios (i.e., rapid onset vs prolonged natural hazards). Second, it presents a novel temporal understanding of neighborhood characteristics by incorporating daytime and nighttime population estimates.

Our core methodology builds upon UrbanPop, a dataset developed by Oak Ridge National Laboratory to produce high demographic and spatial resolution estimates of the United States population [1]. UrbanPop is a microsimulation that combines intercensal data from the ACS PUMS and Longitudinal Employer-Household Dynamics origin-destination employment statistics (LODES), enabling temporal outputs based on PUMS variables.

Using UrbanPop outputs for the 2012-2017 5-year ACS PUMS, we present a case study of Knoxville, TN. We first identify “cohort profiles” based on segments of PUMS samples for each Public-Use Microdata Area (PUMA) in Knoxville. From synthetic populations simulated by UrbanPop, we then perform an ensemble clustering of the cohort profiles to develop daytime and nighttime neighborhood profiles. We conclude with a discussion of our efforts to scale this approach the full United States and its implications for use in a variety of community-level spatial policy interventions.

[1] Morton, A., Nagle, N., Piburn, J., Stewart, R. N., & McManamay, R. (2017). A hybrid dasymetric and machine learning approach to high-resolution residential electricity consumption modeling. In Advances in geocomputation (pp. 47-58). Springer, Cham.

Copyright:  This manuscript has been authored by UT-Battelle, LLC under Contract No. DE-AC05-00OR22725 with the U.S. Department of Energy.  The United States Government retains and the publisher, by accepting the article for publication, acknowledges that the United States Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for United States Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Acknowledgement:  This material is based upon the work supported by the U.S. Department of Energy under contract no. DE-AC05-00OR22725.