Comparison of three agent-based platforms with a simple epidemiological model (WIP)
With the extensive use of agent-based modeling and simula- tion, there are many agent-based platforms available. The ob- jective of this paper is to compare 3 agent-based platforms: NetLogo, Repast, and Cormas based on the results of the simulation obtained from the same set of experimental sce- narios. For this purpose, agent-based SIR model is chosen to study the pattern of the spread of an infectious disease within certain population over time. The methodology for the comparison is to design and implement the same agent-based SIR model in all these platforms, perform numerous experi- ments with the model and compare the outputs using cross- correlation analysis and similarity measure using Manhattan distance. The experimental results show that the agent-based platforms may inherently be biased by design and implemen- tation choices, and they may lead to experimental results that can often be invalidated by other platforms.
Main Authors: | , |
---|---|
Format: | conference_item biblioteca |
Language: | eng |
Published: |
Curran Associates
|
Subjects: | L73 - Maladies des animaux, U10 - Informatique, mathématiques et statistiques, |
Online Access: | http://agritrop.cirad.fr/569559/ http://agritrop.cirad.fr/569559/1/document_569559.pdf |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Summary: | With the extensive use of agent-based modeling and simula- tion, there are many agent-based platforms available. The ob- jective of this paper is to compare 3 agent-based platforms: NetLogo, Repast, and Cormas based on the results of the simulation obtained from the same set of experimental sce- narios. For this purpose, agent-based SIR model is chosen to study the pattern of the spread of an infectious disease within certain population over time. The methodology for the comparison is to design and implement the same agent-based SIR model in all these platforms, perform numerous experi- ments with the model and compare the outputs using cross- correlation analysis and similarity measure using Manhattan distance. The experimental results show that the agent-based platforms may inherently be biased by design and implemen- tation choices, and they may lead to experimental results that can often be invalidated by other platforms. |
---|