Abstract
Agent based modelling is nowadays widely used in transport and the social science. Forecasting population evolution and analysing the impact of hypothetical policies are often the main goal of these developments. Such models are based on sub-models defining the interactions of agents either with other agents or with their environment. Sometimes, several models represent phenomena arising at the same time in the real life. Hence, the question of the order in which these sub-models need to be applied is very relevant for simulation outcomes. This paper aims to analyse and quantify the impact of the change in the order of sub-models on an evolving population modelled using TransMob. This software simulates the evolution of the population of a metropolitan area in South East of Sydney (Australia). It includes five principal models: ageing, death, birth, marriage and divorce. Each possible order implies slightly different results mainly driven by how agents' ageing is defined with respect to death. Furthermore, we present a calendar-based approach for the ordering that decreases the variability of final populations. Finally, guidelines are provided proposing general advices and recommendations for researchers designing discrete time agent-based models.
Highlights
Complex systems characterized by a large number of entities interacting with each others is a very attractive framework to model a large number of phenomena arising in our societies
Our analysis considers the sensitivity of the microsimulation with respect to the choice of the random seeds
This work showed the importance of the order of the models in agent-based modelling, a er having checked the stability against random seeds
Summary
. Complex systems characterized by a large number of entities interacting with each others is a very attractive framework to model a large number of phenomena arising in our societies. Examples of such systems that can involve millions of agents include transportation, social interactions, the spread of contagious diseases and the evolution of populations. The base unit of these models is the agent representing an entity of the population under scrutiny. Each agent is characterised by attributes and behavioural rules mimicking the real entity, and can interact with each other as well as with their environment. Even though the behavioural and interactions rules defined for each agent are typically simple, the resulting emerging behaviour of the system is o en non-linear and di icult to predict
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Journal of Artificial Societies and Social Simulation
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
