Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 87

The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement

Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 86
M O D E L I N G S H O RT- T E R M D Y N A M I C S I N A C T I V I T Y- T R AV E L PAT T E R N S 75 FEATHERS scheduling and are updated each time a schedule is implemented. Learning may involve many mechanisms. Scope First, it is assumed that, as explained earlier, when their activity schedules are implemented, agents will learn This model is part of a wider research program that about the attributes or states of their environment (e.g., involves a number of other Belgian research institutes. travel times) from experiences, which, with respect to the This program aims at examining a series of issues perti- state of a variable, will change the subjective probabili- nent in the development of an activity-based model of ties and hence the agents' beliefs. If the actual situation is travel demand for Flanders, Belgium. For example, new consistent with outcomes perceived as most probable, technology for collecting vehicle data will be explored as uncertainty in beliefs will be reduced, and the individuals well as the application of combined GPSpersonal digi- will be more confident in predicting outcomes on future tal assistant (PDA) technology for collecting activity- occasions. In contrast, if outcomes are contrary to expec- travel data [PARROTS (PDA System for Activity tations, uncertainty will increase and therefore so will Registration and Recording of Travel Scheduling), see difficulty of prediction and perceived value of informa- Bellemans et al. 2005, Kochan et al. 2006]. Feathers tion of future events. Second, in addition to this attribute (Forecasting Evolutionary Activity-Travel of Households learning, it is assumed that agents have an inherent desire and Their Environmental Repercussions) is the acronym to make sense of the world around them. One of the given to the model, which will be based on the current mechanisms involved is identification of the conditions status of the extended Aurora model, as explained ear- that allow them to explain away differences in attributes lier. However, because Aurora to date is largely based on of the environment (condition learning). For example, theory only and on some numerical experiments to assess differences in travel times can be explained in relation to the face validity of the model, further empirical testing day of the week, departure time, weather conditions, an and operational improvement will be required. It is to be accident, and the like. The condition set is not necessar- expected therefore that certain elements will be refined ily constant over time but may grow or shrink. These and that other new elements will be added. The remain- two forms of learning imply that, only after many per- der of this section briefly addresses some of these issues. sonal experiences, agents will have gained sufficient knowledge about their environment. Reality suggests otherwise, and therefore it is assumed that agents also Utility Functions are capable of analogue learning and reasoning: they draw inferences about attributes of certain objects by The core of the models is the S-shaped utility functions analogy with other similar objects. Finally, in addition to (Equations 2 and 3). To date, this shape, which is quite dif- these personal styles of learning, it is assumed that agents ferent from that of other models of activity-time allocation, learn from being part of a social network: they learn by is derived from theory. No specific data to test the shape of word of mouth from members of their social network. the functions and assess its relevance have been collected to Similar Bayesian updating equations (see Arentze and date. Therefore, one of the subprojects is concerned with Timmermans 2006) will be used to estimate these learn- collecting data on how individuals change their activity- ing processes. This is the topic of another project. travel patterns and testing whether the assumed S-shaped utility functions represent such change, or if not, the alter- native functional forms that are required. This project will Impact of Life Trajectory Events also examine and estimate the effect of context variables that influence the maximum utility (Equation 3) that can Above it has been assumed that the household context is be derived. The results will be critical in that, unlike for stationary. However, in reality, the household context other models, it is assumed for this model that utility func- changes over time as a function of life trajectory events, tions are context dependent. Finally, also to be tested is such as a new child, another job, and the like, and this whether the assumed addition of activity and context-spe- change may bring about changes in one or more facets of cific utility functions to represent the overall utility of a the activity agenda and preferences for choice alternatives. daily activity-travel schedule can be corroborated or, if not, The potential relevance and impact of such events in an whether more complex forms are required. activity framework has been explored by van der Waerden et al. (2003a, 2003b) and has led to the formulation of a Bayesian decision network model applied to transport Learning mode choice decisions (Verhoeven et al. 2005a, 2005b). The approach will be further evaluated and extended to It follows from the foregoing that an agent's beliefs about multiple facets of activity-travel patterns in the context of the system with which he or she interacts play a role in Feathers. It constitutes another project in the program.