2013
EMSS 2013 Proceeding
State space analysis for model plausability validation in multi agent system simulation of urban policies
Authors: Miquel Angel Piera, Roman Buil, Egils Ginters
Abstract
Multi-agent models have been increasingly applied to the simulation of complex phenomena in different areas, providing successfully and credible results, however, model validation is still an open problem. The complexity of the stochastic interaction between agents together with a large numbers of parameters can make validation procedures intractable. Particular validation difficulties appear in social science using multi-agent models, when agents are defined as spatial objects to computationally represent the behavior of individuals in order to study emergent patterns arising from micro-level interactions. This paper considers some of the difficulties in establishing verification and validation of agent based models, and proposes the use of colored petri net formalism to specify agent behavior in order to check if the model looks logical and the model behaves logical. Model plausibility is used to express the conformity of the model with a priori knowledge about the process. A proof-of-concept is presented by means of a case study for testing the robustness of emergent patterns through sensitivity analyses and can be used for model calibration. The proposed methodology has been applied in the European Future Policy Modeling project (www.fupol.eu) to create trust and increase the credibility of the agent based models developed to foster e-participation in the design of urban policies by means of simulation techniques.
I3M Scientific Sponsors
I3M Industrial Sponsors
I3M Media Sponsors
