Deriving Decision Rules from Activity Dairies

This paper presents a system for deriving scheduling decision rules from activity diary data. The proposed system represents constraints as well as preferences in determining the sequence of a given set of activities. Rules are organised in a hierarchy and a systematic search procedure is used to optimise the rule hierarchy. A string alignment technique is used to measure the goodness-of-fit of the model in terms of an aggregate distance between observed and predicted schedules. The proposed system is tested based on a large-scale activity diary data set. The results suggest that the optimised rule-set achieves a considerable reduction of aggregate distance. This rule-based approach is suggested as an alternative to existing simultaneous choice models. Potentially, the system is more flexible in adapting schedules to physical or institutional changes in the environment.