Joseph V. Tuccillo
Oak Ridge National Laboratory

James D. Gaboardi
Oak Ridge National Laboratory

Abstract

The ability to produce richly-attributed synthetic populations is key for understanding human dynamics, responding to emergencies, and preparing for future events, all while protecting individual privacy. The Likeness toolkit accomplishes these goals with a suite of Python packages: pymedm/pymedm\_legacy, livelike, and actlike. This production process is initialized in pymedm (or pymedm\_legacy) that utilizes census microdata records as the foundation on which disaggregated spatial allocation matrices are built. The next step, performed by livelike, is the generation of a fully autonomous agent population attributed with hundreds of demographic census variables. The agent population synthesized in livelike is then attributed with residential coordinates in actlike based on block assignment and, finally, allocated to an optimal daytime activity location via the street network. We present a case study in Knox County, Tennessee, synthesizing 30 populations of public K–12 school students \& teachers and allocating them to schools. Validation of our results shows they are highly promising by replicating reported school enrollment and teacher capacity with a high degree of fidelity.

Keywords

activity spaces, agent-based modeling, human dynamics, population synthesis

DOI

10.25080/majora-212e5952-014

Bibtex entry

Full text PDF

Copyright © 2022 Oak Ridge National Laboratory. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.