A primer on data-driven modeling of complex social systems

TL;DR

This tutorial introduces data-driven modeling techniques for complex social systems, emphasizing different modeling approaches, challenges, and applications like election dynamics and pedestrian movement.

Contribution

It provides a comprehensive overview of modeling methods, differences, and challenges in understanding complex social systems through data-driven approaches.

Findings

Models help predict group-level features from individual interactions.

Comparison of statistical, mathematical, static, dynamic, spatial, non-spatial, discrete, continuous, phenomenological, and mechanistic models.

Discussion of challenges in model building, calibration, and data integration.

Abstract

Traffic jams on roadways, echo chambers on social media, crowds of moving pedestrians, and opinion dynamics during elections are all complex social systems. These applications may seem disparate, but some of the questions that they motivate are similar from a mathematical perspective. Across these examples, researchers seek to uncover how individual agents -- whether drivers, Twitter accounts, pedestrians, or voters -- are interacting. By better understanding these interactions, mathematical modelers can make predictions about the group-level features that will emerge when agents alter their behavior. In this tutorial, which is based on the lecture that I gave at the 2021 American Mathematical Society Short Course, I introduce some of the terms, methods, and choices that arise when building such data-driven models. I discuss the differences between models that are statistical or mathematical, static or dynamic, spatial or non-spatial, discrete or continuous, and phenomenological or mechanistic. For concreteness, I also describe models of two complex systems, election dynamics and pedestrian-crowd movement, in more detail. With a conceptual approach, I broadly highlight some of the challenges that arise when building and calibrating models, choosing complexity, and working with quantitative and qualitative data.