Sociodynamic Discrete Choice Applied to Travel Demand: Multi-Agent Based Simulation and Issues in Estimation

This paper discusses a multi-agent based model of binary choice behavior with interdependence of decision-makers' choices. Analytical results established by other authors are briefly summarized where agent heterogeneity is not explicitly treated. Next the well-known Erdos-Renyi network class is considered to introduce agent heterogeneity via an explicit local interaction structure. Then the model is applied in an example of intercity travel demand using empirical data to introduce individual agent heterogeneity beyond that induced by the local interaction structure. Studying the long run behavior of more than 120,000 multi-agent based simulation runs reveals that the initial estimation process can be highly sensitive to small variations in network instantiations. We show that this is an artifact of two issues in estimation, and highlight particular attention that is due at low network density and at high network density. Limitations in the present work are summarized and suggestions for future research efforts are outlined.