Development of an Activity-based Microsimulation Model for Daily Travel based on Tour Concept

Abstract

This research aims to develop an activity-based travel demand model applying to the “tour concept” which was proposed by Bowman et al. The neural network model (NN model) is used to estimate the structure of all models, rather than the practical disaggregate logit model. A questionnaire survey was carried out for gathering activity diary data in the region of Niitsu, Niigata. Three different kind of models, a simultaneous type, and types of sequence without feedback and with feedback, are estimated using the NN network and their microsimulation models are developed to predict each person's travel. The type of sequence model with feedback can achieve higher Hitting Ratio than the other two types. Among the sub-models of sequence type, destination choice is most difficult to hit. The research shows it is important to consider inter-relationship and influences among travel choice components.