Experiences with developing ALBATROSS: a learning-based transportation oriented simulation system

This paper discusses some experiences with the development of ALBA TROSS, a rule-based system for predicting transport demand, currently under development for the Dutch Ministry of Transport, Public Works and Water Management. The model belongs to the class of activity-based models, implying that it attempts to predict which activities are conducted where, when, with whom, for how long, and the transport mode involved. In principle, this increased complexity allows one to predict the impact of urban, and transport policies and institutional change on activity pattems and hence transport demand, but this increased complexity also involves new theoretical, and methodological challenges and problems of data collection. Some of these challenges are briefly discussed in this paper. In particular, the conceptualisation of activity behaviour, the derivation of choice heuristics from diary data, the development of appropriate goodness-of-fit measures and the problem of data quality are discussed