Travel demand generation using Bayesian Networks: an application to Switzerland

Thanks to their ability simulate the travel behavior at individual scale, agent-based models have gained popularity over last years. These are data-intensive, with regards transport supply and demand. In particular, a detailed description of population its is required. Bayesian Networks (BNs) directed acyclic graphs representing joint probability distributions. They recently been employed for synthesis daily activity patterns generation in studies showing that BNs effectively capture causality links existing between variables easily interpretable. Moreover, given flexible structure, can be adapted situations which data from various sources combined. this paper, our goal estimate BN both pattern Switzerland. We evaluate performance approach compared statistical matching algorithm using aggregated disaggregated metrics. we show understanding dependency structure linking characteristics mobility key generate representative synthetic agents patterns. This study contribution towards development interpretable, behaviorally rich demand models.