Calibrating an Agent-Based Model of the Ambient Population using Big Data

Authors: Tomas Crols*, University of Leeds, Nick Malleson, University of Leeds
Topics: Spatial Analysis & Modeling
Keywords: agent-based modelling, big data, ambient population
Session Type: Paper
Day: 4/12/2018
Start / End Time: 8:00 AM / 9:40 AM
Room: Bayside A, Sheraton, 4th Floor


While many datasets and methodologies are currently available to map the night-time residential population, less research has focused on the ambient population, the actual location of people throughout the day. Live forecasts of the ambient population in urban areas could have applications in different fields that need to know how many people are at risk or in need of something at any time. These include crime science, transport research, or modelling exposure to air pollution. Recently, human movements can be tracked thanks to a diverse range of ‘big’ datasets having a high spatial and temporal resolution. Examples are mobile phone locations, footfall cameras, geolocated social media updates or transport smart cards. In this presentation, we discuss the development of an agent-based model of the ambient population at an individual level in an urban environment. Ultimately, the goal is to calibrate the model with different big data streams in real time by using dynamic data assimilation techniques. At the moment, we have developed a model of recurring activities of individuals (commuting, shopping, leisure, etc.) that we calibrated with hourly Wi-Fi sensor footfall data in the town centre of Otley, West Yorkshire, UK. The dataset captures mobile phones passing by at different well-chosen crossings and buildings.

Abstract Information

This abstract is already part of a session. View the session here.

To access contact information login