Agents in Space: Validating ABM-GIS Models

The purpose of this paper is to spatially validate an agent-based predictive analytics model of energy siting policy in a techno-social space. This allows us to simulate the multitude of human factors at each level (e.g. individual, county, region, and so on). Energy infrastructure siting is a complex and contentious process that can have major impacts on citizens, communities, and society as a whole. Furthermore, the process is sensitive to varying degrees of human input, of differing complexity, at multiple levels. When it comes to validating ABMs, the virtual cornucopia of techniques can easily confuse the modeler. As useful as historical data validation is, it seems to be underutilized, most likely due to the fact that it is hard to find data suitable data for many models. For the purpose of In-Site, historical data availability is excellent due to Environmental Impact Assessments (EIA) providing us with citizen and community based organization (CBO) preferences, and regulatory decisions being public. For the model, citizen and CBO preferences were decided by coding comments on the EIA procedure so as to allow for quantitative analysis, and then geocoding the locations of the commenters. The end results of this is that, we can literally overlay our simulation results with the actual, real world, results of the historical project. This will allow for a high degree of confidence in the validation procedure, as well as the ability to deal with the complexity of the networks of human interactions.