Analysis of activity choice: The role of activity attributes and individual schedules

Abstract

Activity-based approaches have taken hold in transportation research over the last several decades. The foundation of the activity-based approach is to view travel as a result of our activity choices and scheduling decisions. Therefore, better understanding of activity choice, planning time horizons, and activity attributes will lead to more accurate demand forecasts. This dissertation extends the current activity choice modeling framework by incorporating the characteristics of the individuals’ schedules, planning time horizons and focusing on the salient attributes of the activities.

This study consists of three parts which are linked to one another by their conceptual and empirical findings. The first part identifies the determinants of the planning time horizons - defined as when people decide about performing their activities. Several household and individual characteristics, and activity attributes are tested for their association with planning times. The activity attributes which have significant impacts on the planning time horizons of the activities are used in the second part for generating new activity groups. The second part clusters activities based on their salient attributes, such as duration, frequency, number of involved people and flexibilities, rather than their functional types (work, leisure, household obligations, etc.) and creates activity groups such as "long, infrequent, personally committed activities", "quick, spatially fixed, temporally flexible activities" etc. The activity groups generated in this part inform the activity choice modeling structure developed in the third part. The main analytical techniques used in this research are the Principal Components Analysis (PCA) and discrete choice models. PCA is used to define the new activity groups. The analysis of the planning time horizons and activity choice are performed by mixed logit models.

The model results reveal the significant relationships between socio-demographics, temporal characteristics, travel, and characteristics of the schedules on activity choice. The findings of these models could be integrated in the activity choice modules of the existing activity-travel simulation models by either applying the comprehensive model (which may face limitations due to the availability of data) or integrating the findings of the models in the decision rules.