Abstract
AbstractBetter understanding of the behaviour of agent-based models, aimed at embedding them in the broader, model-based line of scientific enquiry, requires a comprehensive framework for analysing their results. Seeing models as tools for experimenting in silico, this chapter discusses the basic tenets and techniques of uncertainty quantification and experimental design, both of which can help shed light on the workings of complex systems embedded in computational models. In particular, we look at: relationships between model inputs and outputs, various types of experimental design, methods of analysis of simulation results, assessment of model uncertainty and sensitivity, which helps identify the parts of the model that matter in the experiments, as well as statistical tools for calibrating models to the available data. We focus on the role of emulators, or meta-models – high-level statistical models approximating the behaviour of the agent-based models under study – and in particular, on Gaussian processes (GPs). The theoretical discussion is illustrated by applications to the Routes and Rumours model of migrant route formation introduced before.
Publisher
Springer International Publishing
Reference66 articles.
1. Andrianakis, I., Vernon, I. R., McCreesh, N., McKinley, T. J., Oakley, J. E., Nsubuga, R. N., Goldstein, M., & White, R. G. (2015). Bayesian history matching of complex infectious disease models using emulation: A tutorial and a case study on HIV in Uganda. PLoS Computational Biology, 11(1), e1003968.
2. Angione, C., Silverman, E., & Yaneske, E. (2020). Using machine learning to emulate agent-based simulations. Mimeo. arXiv. https://arxiv.org/abs/2005.02077. (as of 1 August 2020)
3. Banks, D., & Norton, J. (2014). Agent-based modeling and associated statistical aspects. In International Encyclopaedia of the Social and Behavioural Sciences (2nd ed., pp. 78–86). Oxford University Press.
4. Bijak, J. (2010). Forecasting international migration in Europe: A Bayesian view. Springer.
5. Bijak, J., & Bryant, J. (2016). Bayesian demography 250 years after Bayes. Population Studies, 70(1), 1–19.