Global sensitivity/uncertainty analysis for agent-based models

Authored by Vladimir A. Fonoberov, Maria Fonoberova, Igor Mezic

Date Published: 2013-10

DOI: 10.1016/j.ress.2013.04.004

Sponsors: United States Air Force

Platforms: C++ Microsoft Visual Studio

Model Documentation: Other Narrative Pseudocode Mathematical description

Model Code URLs: Model code not found

Abstract

Agent-based models simulate simultaneous actions and interactions of multiple agents, in an attempt to re-create and predict the appearance of complex phenomena. We propose to use global sensitivity analysis as a tool for analyzing and evaluating agent-based models. A general approach for applying the global sensitivity analysis to agent-based models is presented and tested on the example of a sociocultural agent-based model we developed earlier [45]. We identify the most significant parameters in the model and uncover their contributions to the outputs of interest. Methodology of model reduction for agent-based models is discussed and demonstrated for the aforementioned model. (c) 2013 Elsevier Ltd. All rights reserved.

Tags

Agent-based model global sensitivity analysis Uncertainty analysis Derivative-based global sensitivity Model reduction Variance-based global sensitivity