A Markov Chain Monte Carlo Approach for Estimating Daily Activity Patterns

Determining the purpose of trips brings is a fundamental information to evaluate travel demand during the day and to predict longer-term impacts on the population’s travel behavior. The concept of tours is the most suited to consider the value of a daily scheduling of individuals and travel interdependencies. However, the meticulous care required for both collecting data of high quality and interpret results of advanced demand models are frequently considered as major drawbacks. The objective of this study is to incorporate into a standard trip-based model some inherent concepts of activity-based models in order to enhance the representation of travel behavior. The main focus of this work is to infer, employing utility theory, the trip purpose of a population, at a zonal level. Making use of Markov Chain Monte Carlo, a set of parameters is estimated in order to retrieve tour-based primitives of the demand. The main advantage of this methodology is the low requirements in terms of data, as no individual information are used, and the good interpretation of the model. Estimated parameters of the priors set a utility-based probability function for departure time, which allows to have a dynamic overview of the demand. In order to account for the tour consistency of travel decisions, a duration constraint is added to the model. The proposed model is applied to the region of Luxembourg city and the results show the potential of the methodologies for dividing an observed demand based on the activity at destination.