Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 78

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 77
NEXT STEPS 65 methods are introduced, the practicing community will A second obstacle relates to the limited experience face many uncertainties relating to needed data, model that we currently have with the use of activity models development procedures, and application software. and other advanced techniques. Experience is limited at There is also uncertainty about how consultants should MPOs and in the consultant community. For widespread be selected and used in developing advanced techniques. application of these techniques, we need widespread dis- It will take time for the practitioners to become educated semination of information on experiences with the use of and comfortable with new methods. different models. The TPB is planning to conduct a survey of 10,000 The Metropolitan Transportation Commission households in the coming year to obtain current travel (MTC) held a peer-review panel funded through the behavior data. We plan to use the survey data to support Travel Model Improvement Program in December 2004. the update of our trip-based models, and, in the longer Our next activity was a model specification and training term, to support the development of activity-based mod- study. We are in the process of reviewing our specifica- els. While we can discuss the advantages of activity mod- tion plan. We have gone through some very intensive in- els over conventional models, I still want to stress the house training over the past few months. The MTC staff point that no model will operate well if the inputs to the will estimate the activity-based model. We will use con- model are of poor quality. It is important to remember sultant assistance with the specification plan, the over- that the inputs to the model are critical to arriving at sight of staff estimation of the models, and the software realistic results. We should ensure that adequate construction after we are done with the estimated mod- resources are focused not only on modeling improve- els. We are using a work-sharing approach to these activ- ments, but also on striving to improve input data to the ities. I am very excited about the process. model. The training effort focused on multimodal logit mod- els, nested logit models, destination choice models, esti- Chuck Purvis mation of daily activity patterns, interhousehold interaction models, and tour departure and duration T he papers and presentations have been interesting and informative. I would give the best paper award to Ram Pendyala and Chandra Bhat for their paper, "Vali- choice models. This approach gives MTC staff a pride of ownership and an excellent understanding of the mod- els. This approach may not work for all MPOs, but for dation and Assessment of Activity-Based Travel Demand us, estimating the models in house and working in part- Modeling Systems." I also found the information on the nership with consultants is an excellent way to imple- work under way at the Denver Regional Council of Gov- ment a new set of travel behavior models. ernments and the Sacramento Area Council of Govern- I also have a concern over whether we should refresh ments to be very interesting. We can all learn from their our trip-based models. Our approach of using staff experiences as we move forward with model estimation rather than consultants to implement activity-based with the next generation of activity-based models. I also models will probably take longer than relying totally on learned about clock-time savings that one can accrue by consultants. I think that in the end we will have a better freezing an activity-based model process after certain model, a better process, and a better-trained staff. steps in the model application. One topic I would like to learn more about is the level Aichong Sun of uncertainty in our activity-based models. We need to present a range of values to represent this uncertainty. I would also like to hear more about the validation of the synthetic population generation component. We do have I appreciate the opportunity to provide a perspective from a medium-sized MPO. Voters in Pima County, Arizona, recently approved an increase in the sales tax multiple years of public use microdata sample data and that is dedicated to transportation projects in the county. we need to thoroughly test the population synthesizers The sales tax is expected to generate approximately $2.1 using past census data and the new American Com- billion over the next 20 years. muters Survey that should be available this fall. We use a conventional four-step travel model at the I think training and education is the number one Pima Association of Governments. We just completed a obstacle to advancing the use of activity models and model update with the assistance of consultants. Three other techniques. We also need a summary of the issues components were added to the model as part of this related to the non-market-segmented trip-based update. These components are a household model, a models, the market-segmented trip-based models, and time-of-day model, and a mode choice model. the microsimulated models. We need a discussion of The outputs of the household model include house- the issues related to ecological fallacies associated hold income, automobile ownership, and the number of with moving from a naive segmentation to the full workers. Previously, we could only model 24-hour traf- microsimulation. fic volumes. The time-of-day model allows us to model