Activity-travel pattern choice modeling within space-time prisms

Abstract

Route choice set generation and route choice modeling have been studied for years, but most focus on the trip level and rarely address activity-travel patterns (ATPs) in activity-based scheduling frameworks. This study proposes an integrated framework to generate ATP choice sets under space-time constraints and model ATP choices while considering overlapping among alternatives. The model framework generates ATP choice sets with flexible activity sequences using a link penalty approach and defined ATP utility functions of MNL and MNL-modification models to account for the spatial and temporal overlapping. Utilizing GPS trajectories, the simulation and estimation results of the ATP choice set generation and ATP choice modeling demonstrate the validity of the model framework. The ATP choice model comparison reveals the goodness-of-fit of capturing the correlations among the alternatives and individual heterogeneity.