Data-Driven ABM Methodology and Applications

Lead Investigators

  • Hamdi Kavak
  • Jose Padilla (Advisor)


Student and Senior Collaborators

  • Saikou Diallo (Collaborator faculty 2014-2018)
  • Christopher Lynch (Collaborator student 2014-2018)
  • Olcay Sahin (Collaborator student 2013-2014)
  • Brit Nicholson (Collaborator student 2013-2014)


Project Dates

2013-2018


Methodology

We have recently witnessed the proliferation of large-scale behavioral data that can empirically develop agent-based models (ABMs). However, current practices in agent-based modeling do not offer a structured approach to creating agents or their different computational parts directly from data. This project illustrates an agent-based modeling approach that leverages individual-level data to generate agent behavioral rules and initialize agent attribute values. You will first see a few sketches of this approach below, followed by some related applications.


Use of Social Media Data in ABMs and Applications

As a pre-cursor to the data-driven approach, we studied how social media could be used to support agent-based modeling efforts. In the paper cited below (Padilla et al. 2014), three example usages are shown:

  1. Using large data set processing and sentiment analysis to identify preferences of a population.
  2. Using agents with machine learning capabilities to learn mobility patterns from individuals’ history.
  3. Identifying preferences and communication patterns based on graph analysis (agent relation).
    The first item in this list was applied in a healthcare-related simulation setting, cited below as Padilla et al. (2015). Later, a complete version of the approach is formalized, two example applications using different datasets were developed, and the effectiveness of the approach is showcased in Kavak (2019).


Publications & Presentations

  • A Data-Driven Approach for Modeling Agents
    H. Kavak
    Old Dominion University, 2019
    [Dissertation] [BibTex]
  • Big Data, Agents, and Machine Learning: towards A Data-Driven Agent-based Modeling Approach
    H. Kavak, J.J. Padilla, C.J. Lynch, and S.Y. Diallo
    51st Annual Simulation Symposium, Baltimore, MD, April 15-18, 2018, doi: 10.22360/SpringSim.2018.ANSS.021
    [Paper] [BibTex]
  • Human Mobility Simulation Framework Using Big Data and Agents (Best presentation award)
    H. Kavak and J.J. Padilla
    2016 Spring Simulation Conference, Poster Session and Student Colloquium, doi:10.13140/RG.2.1.4179.7522
    [Poster]
  • Leveraging Social Media Data in Agent-based Simulations.
    J.J. Padilla, S.Y. Diallo, H. Kavak, O. Sahin, and B. Nicholson.
    47th Annual Simulation Symposium, Tampa, FL, April 13-16, 2014
    [Paper] [BibTex]
  • Semi-Automated Initialization of Simulations: An Application to Healthcare.
    J.J. Padilla, S.Y. Diallo, H. Kavak, O. Sahin, J.A. Sokolowski and R.J. Gore
    The Journal of Defense Modeling and Simulation, 2015, doi:10.1177/1548512915570143
    [Paper] [BibTex]


Funding

  • The Office of the Assistant Secretary of Defense for Research and Engineering (OASD(R&E)) under agreement number FAB750-15-2-0120.

Last updated on Jan 27, 2022.