University of California Transportation Center

Activity-based approaches are perhaps the most promising alternative to the current travel forecasting methodology. This dissertation first presents a pattern generation model that can serve as a link between activity and trip-based methodologies. The model uses a clustering approach to identify groups of similar activity-travel behavior and relates them to household socioeconomic attributes. Minimally, the pattern generation model is then expanded to serve as the core component of a proposed activity-based microsimulation model that constructs complete origin-destination tables using a wholly activity-based approach. The techniques developed provide due diligence to the complex nature of activity-travel behavior in terms of spatial and temporal constraints, household interactions, and the derived nature of such behavior. A successful application of the expanded model is outlined using data from the 1994 Portland activity-travel survey.