Geoscience Reference
In-Depth Information
9.6 Guidance for Students of AB Modelling .............................................................................. 228
9.6.1 Before Starting Development of the Long-Living Version of Your Model .............. 228
9.6.2 When Developing and Coding the Simulation ......................................................... 228
9.6.3 When Investigating Your Model Theoretically ........................................................ 229
9.6.4 When Applying Your Model to the Real World ....................................................... 230
9.7 Conclusions ........................................................................................................................... 230
Appendix 9A Popular Topics in AB Modelling .......................................................................... 231
References ...................................................................................................................................... 233
ABSTRACT
Agent-based (AB) modelling is a leading methodology of dynamic GeoComputation that aims at
making a direct connection between the knowledge on human behaviour and decision making and
the dynamics of geographic systems. An AB model implements spatial and non-spatial aspects of
the behaviour of individuals and institutions within an explicitly represented geographic environ-
ment. In this way, it enables investigation of the dynamics and evolution of geographic systems as
driven by humans. An AB model is a sort of a virtual world populated by avatars whose behaviour
reflects decades of scientific research in social and physical geography, economics and behavioural
science. In this way, researchers can directly study if and how humans can change themselves,
nature and society.
This chapter presents a methodology for AB modelling and focuses on operational issues: the
choice of agents, the representation of relationships between them, the formulation and formalisa-
tion of the agent behavioural rules, a study of the AB model dynamics and an application of the
AB models to real-world problems. The approach is illustrated with the help of two AB models:
the famous Schelling model of ethnic residential dynamics and PARKAGENT, a model of parking
search in the city. The chapter finishes with a list of recent developments in the field of AB model-
ling and a set of guidelines aimed at students wanting to adopt an AB modelling approach.
9.1 ANIMATED AGENTS AND NON-ANIMATED OBJECTS:
THE PILLARS OF GEOGRAPHIC SYSTEMS
We live in a 4D spatio-temporal world and GeoComputation (GC) captures the pieces of this world at
certain
geographical resolutions
, which start at the level of a single human and reach up to encompass
the entire Earth. The essence of GC is a quantitative analysis of these pieces. Often, we are interested
in understanding an averaged image of the world. In this case, we apply static methods of analysis
that account for the spatial dimensions of the phenomenon but ignore their dynamics. However,
when our interest lies in modelling the dynamics of spatial phenomena, then such static analyses
become inadequate. This chapter presents
Agent-Based
(A B)
modelling
* as a natural framework
for investigating the
spatio-temporal
dynamics
of geographic systems. Among other approaches
to modelling natural phenomena, geographical AB modelling is unique in one respect: the
agents
in an AB model represent, explicitly or implicitly, human individuals and institutions who imitate
the decision-making behaviour of humans. In this way, geographical AB modelling considers the
dynamics of geographic systems as driven by humans and their decisions. The agents in ecological
AB models can reflect animals and animal behaviour, while in computer science, the software agents
are autonomous computer programs. Geographical AB modelling focuses on human-driven systems.
This chapter focuses on AB modelling as an increasingly popular GC approach to formalis-
ing and studying complex geographic systems. The reader is assumed to be a regular user of GIS
(geographic information systems) and to have some knowledge of relational databases and object-
oriented programming.
*
Here and in the following, we use italics when presenting a notion for the first time.
