UNCERTAINTY ANALYSIS OF THE RULE-BASED ALBATROSS MODEL

Uncertainty is difficult to judge in complex large scale computational process models, which consist of a large number of decision rules applied in a sequential fashion. The purpose of this paper is to examine the issue of model uncertainty in one such model: Albatross. Since Albatross uses Probabilistic Action Assignment Rules to assign decision alternatives to cases, the model has some inherent uncertainty. In this paper, we will describe the results of an uncertainty analysis of the Albatross model. The model was run 50 and 100 times for 10% of a synthetic population, created for the city of Rotterdam, The Netherlands. Based on these runs, the coefficient of variation and the number of runs required to obtain reliable results were calculated for some key travel indices. The uncertainty in these indices is compared for different choice facets scheduled at different stages in the sequence of decision tables.