Title of Paper: “Modeling and Simulating Demographic Dynamics within an Agent-based Integrated Urban Model: Application of Machine Learning Techniques”
Authors: Khalil, M.A., Fatmi, M.
Description: The demographic characteristic of any region directly affects its transportation demand and energy consumption, among others. Most of the existing demographic components rely on logit and rule-based models, which can be inflexible to capture the decision complexities. In this article, we developed a machine learning (ML) demographic dynamics (DYx) module within an agent-based integrated urban model (IUM). In principle, our DYx includes aging, childbirth, death, education change, children leaving their parents’ homes, getting married, divorce, job change, and in-and-out migration. The results show that ML models outperform the conventional logit models by 4%-15%. Furthermore, we utilized an explainable AI technique (xAI) to improve the transparency of the ML models by understanding the influence of independent and dependent variables. The article also presents the microsimulation results of the DYx deployed within an IUM for the 100% population of the Central Okanagan region of British Columbia for 2011-2016 where the preliminary validation exercise suggests that the simulation results are reasonably satisfactory.
Presentation Location: Hall A, Convention Center
Presentation Time: January 9th, 2023 10:15AM- 12:00PM EST
Comments are closed, but trackbacks and pingbacks are open.