Generating macroscopic, purpose-dependent trips through Monte Carlo sampling techniques

While estimating origin-destination (OD) demand flows usually requires a large amount of data, nowadays a key issue in traffic engineering is to estimate the trip purpose while protecting user privacy. The aim of this work is to derive from macroscopic and aggregate information the temporal distribution for the production of each traffic zone of a system, with a trip-purpose specification. We suggest different procedures for estimating the production factors, which are based on the precision level of the available information. If time-dependent demand data is available, the production factor can be estimated through a simple Monte Carlo simulation model. Otherwise, a Markov Chain Monte Carlo (MCMC) approach is proposed to approximate a set of functions that describe the production of purpose-specific trips with regard to one specific zone along the day. This algorithm requires a lower level of input information and computes the likelihood with regard to the number of generated and attracted trips. Application of the models is shown using available real data collected through a one-week travel diary within the area of Ghent, Belgium.