Agent-Based Simulation Modeling: Definitions and a Methodological Proposal

Agent-Based Simulation Modeling: Definitions and a Methodological Proposal

Marco Valente
Copyright: © 2016 |Pages: 12
DOI: 10.4018/978-1-4666-9770-6.ch004
OnDemand:
(Individual Chapters)
Available
$37.50
Current Special Offers
No Current Special Offers
TOTAL SAVINGS: $37.50

Abstract

Computer simulations are a powerful tool for scientific research, but lack an accepted methodology for their use, and consequently their results are generally received with skepticisms. This chapter proposes a methodological approach allowing to formally unify the treatment of “traditional” quantitative phenomena with that of phenomena from economics or biology that prevent a universal adoption of data-centered methods. We propose to adopt the explanation as the basic unit of knowledge, which is able to cover all possible cases. From this assumption, we can derive the conclusion that simulation models fail to deliver their full potential as scientific investigative tool because their implementations lack crucial details on the intermediate steps producing simulation results.
Chapter Preview
Top

2 Describing Vs. Explaining

When asking modelers how they believe their models should be assessed, the standard answer is related to the degree if adherence of the model results to observation of reality: the closer the model to observation, and the larger the number of real-world features captured by the model, the better fares the model against competing alternatives. In short, a model should be assessed in terms of validation as performed in physical sciences. Simulation models present specific difficulties, but a growing literature is starting to define how validation should be performed for this kind of models (Windrum, Fagiolo & Moneta 2007), and authors are increasingly devoting efforts in supporting the reliability of their results (Dosi, Fagiolo & Roventini 2010, Valente 2012).

Complete Chapter List

Search this Book:
Reset