| Abstract | With the extensive use of agent-based modeling and simulation these days, many agent-based
platforms are available. The objective of this research study is to compare three agent-based
platforms: NetLogo, Repast, and Cormas based on the results of the simulation obtained
from the same set of experimental scenarios. For the research 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 are to design and implement
the same agent-based SIR model in all these platforms, perform numerous experiments with
the model and compare the outputs obtained from them using cross-correlation analysis and
similarity measure using Manhattan distance. We also compare these platforms based on
the selection criteria like popularity index of agent-based simulation platform, documentation,
modeling speed, ease to use, programming efforts, programming language, simulation
speed, GUIs, ease of installation, and license. From the experimental tests, we conclude that
all agent-based platforms do not give the similar results. We found that there are similarity
between the platforms NetLogo and Cormas but dissimilarity between the platforms Repast
and Cormas. In the decreasing order of priority, the platforms for beginner modeler are Net-
Logo, Cormas and Repast. Agent-based platforms are not for making accurate predictions
or forecasting because we cannot trust the results. |