Practical Points for the Software Development of an Agent-Based Model of a Coupled Human-Natural System

Authored by Santiago L Rovere, Michael J North, Guillermo P Podesta, Federico E Bert

Date Published: 2016

DOI: 10.1109/access.2016.2592418

Sponsors: No sponsors listed

Platforms: Java

Model Documentation: UML Other Narrative Flow charts

Model Code URLs: https://www.comses.net/codebases/3872/releases/1.0.0/

Abstract

Modeling complex natural and human systems to support policy or management decision making is becoming increasingly common. The resulting models are often designed and implemented by researchers or domain experts with limited software engineering expertise. To help this important audience, we present our experience and share lessons learned from the design and implementation of an agent-based model of agricultural production systems in the Argentine Pampas, emphasizing the software engineering perspective. We discuss the model's design including the model classes; the activity diagram, and data flow; the package and folder layout; the use of design patterns; performance optimization; initialization approaches; the analysis of results; and model measurement, validation, and verification.
Tags
Simulation Management Market Land-use change Economics Sensitivity-analysis Agricultural systems Metrics Aspiration-level Argentine pampas