An integrated framework for forecasting travel behavior using Markov Chain Monte-Carlo simulation and profile Hidden Markov Models

Recent advances in agent-based micro-simulation modelling have further underlined the importance of a thorough full synthetic population procedure to guarantee a correct characterization of the true population. We propose an integrated approach including Markov Chain Monte Carlo (MCMC) simulation and profiling based methods to catch the complexity and the diversity of agents of the true population through representative micro samples. The population synthesis method is capable of building the joint distribution of a given population with its corresponding marginal distributions (e.g. age, gender, socio-professional status etc.) using complete, partial conditional probabilities or both of them at the same time. Particularly, the estimation of socio-demographic variables and characterization of daily activity-travel patterns are included within the framework. Data stemming from the 2010 Belgian Household Daily Travel Survey (BELDAM) are used to calibrate the modelling framework. We illustrate that this framework catches in an efficient way the behavioral heterogeneity of travelers. Furthermore, we show that the proposed framework is adequately adapted to build large scale micro-simulation scenarios of transportation and urban systems.