Paper and Slides: "Analysing a Complex Agent-Based Model Using Data-Mining Techniques"

Analysing a Complex Agent-Based Model Using Data-Mining Techniques 
By Claire Little, Bruce Edmonds, Laurence Lessard-Phillips, and Ed Fieldhouse

Presented at Social Simulation 2014, Barcelona, September.

A complex “Data Integration Model” of voter behaviour is described. However it is very complex and hard to analyse. For such a model “thin” samples of the outcomes using classic parameter sweeps are inadequate. In order to get a more holistic picture of its behaviour data- mining techniques are applied to the data generated by many runs of the model, each with randomised parameter values.

Paper at: http://cfpm.org/aacabm/analysing a complex model-v3.4.pdf
Slides at:  http://www.slideshare.net/BruceEdmonds/analysing-a-complex-agentbased-model-using-datamining-techniques

Slides of my talk at the 2014 ESSA summer school: "Winter is coming! - how to survive the coming critical storm and demonstrate that social simulations work"

 A talk at the 2014 European Social Simulation Association summer school, at UAB in Barcelona 8th sept 2014

The talk covers some of the symptoms of hype in social simulation and argues that it needs to be more careful and rigourous. In particular that the (current) purpose of a simulation needs to be distinguished between theoretical, explanatory or predictive. Each having their own critieria.

http://www.slideshare.net/BruceEdmonds/winter-is-coming-38816017