Using surrogate models to calibrate agent-based model parameters under data scarcity

Social Simulation is one of the most prominent uses of Multiagent Systems, but it requires the costly task of fitting parameters to assure the credibility of the model. As, to date, there is no con-sensus on how to calibrate parameters of agent-based models, we have investigated other knowledge domains to develop an efficient method for this task. Our proposal is based on the definition of a surrogate model, that reduces search space dimension. We have tested our method in the housing market scenario, using real data. We achieved satisfactory results, that corroborate the idea that it is important to reduce the search space for an efficient parameter calibration.