Spatial agent-based models in urban applications

No Thumbnail Available
Journal Title
Journal ISSN
Volume Title
Insinööritieteiden korkeakoulu | Master's thesis
Date
2014-01-27
Department
Major/Subject
Geoinformation Technology
Mcode
IA3002
Degree programme
Master’s Programme in Geoinformatics
Language
en
Pages
52
Series
Abstract
Agent-based modelling (ABM) provides the simulation of complex emergent behavior of the system by simulation of individual behavior of agents that act within it. The purpose of this thesis was to investigate the role of space in agent-based models. Another aim was to find how spatial data can be modeled using agent-based approach in urban applications. Finally, to test the theoretical positions of the thesis, the creation of simplified urban agent-based application was intended. The work was based on the literature review that covered the main areas of implementation of ABM with emphasis on urban studies. The findings of review were used for investigation of theoretical research questions and in creation of the simple urban agent-based model dedicated to building placement. In the scope of the thesis the main approaches to modelling of spatial phenomena and their difference were demonstrated. The spatial relationships in geoinformatics, urban studies and ABM were identified. The results of the thesis revealed that space is presented in ABM like environment and it endows the agents the property to set metrical and topological spatial relationships of different types. Finally, the approach to dealing with spatial agent-based model and its results was demonstrated. It covered all the stages including the formulation of the rules, setting the model, proposal of testing experiments, suggestion of methods for analysis of simulation results and model validation. The results of experiments demonstrated sensitivity of the model to the shape of environment and the need of formulation precise comprehensive rules that take into account spatial autocorrelation between events. The potential of agent-based modeling identified in this work needs further study and should be used to create more sophisticated models, in particular in the field of rule-based urban planning.
Description
Supervisor
Virrantaus, Kirsi
Thesis advisor
Nikander, Jussi
Keywords
agent-based model, space, spatial relationships, spatial modelling
Other note
Citation