Methodology to model activity participation using longitudinal travel variability and spatial extent of activity

• Main Author: Elango, Vetri Venthan
• Other Authors: Guensler, Randall L.
• Format: Others
• Language: en_US
• Published: Georgia Institute of Technology 2016
• Subjects: Activity participation model; Structural equations model; Travel behavior variability; Activity space; Trip purpose; GPS; Data processing
• Online Access: http://hdl.handle.net/1853/54290

Macroscopic changes in the urban environment and in the built transportation infrastructure, as well as changes in household demographics and socio-economics, can lead to spatio-temporal variations in household travel patterns and therefore regional travel demand. Dynamics in travel behavior may also simply arise from the randomness associated with values, perceptions, attitudes, needs, preferences and decision-making process of the individual travelers. Most urban travel behavior models and analysis seek to explain variations in travel behavior in terms of characteristics of the individuals and their environment. Spatial extents and temporal variation in an individual’s travel pattern may represent a measure of the individual’s spatial appetite for activity and the variability-seeking nature on his/her travel behavior. The objective of this dissertation effort is to develop a methodology to predict activity participation using revealed spatial extents and temporal variability as variables that represent the spatial appetite and variability-seeking nature associated with individual household. Activity participation is defined as a set of activities in which an individual or household takes part, to satisfy the sustenance, maintenance and discretionary needs of the household. To accomplish the goals of the dissertation, longitudinal travel data collected from the Commute Atlanta Study are used. The raw Global Positioning Systems (GPS) data are processed to summarize trip data by household travel day and individual travel day data. A methodology was developed to automatically identify the activity at the end of each trip. Methods were then developed to estimate travel behavior variability that can represent the variability-seeking nature of the individual. Existing methods to estimate activity space were reviewed and a new Modified Kernel Density area method was developed to address issues with current methods. Finally activity participation models using structural equation modeling methods were developed and the effects of the variability-seeking nature and spatial extent of activities were applied to the models. The variability-seeking nature was presented in the activity participation model as a latent variable with coefficient of variation of trips and distance as indicator variables. The dissertation research found that inclusion of activity space variables can improve the activity participation modeling process to better explain travel behavior.