Cover Image

Not for Sale

View/Hide Left Panel
Click for next page ( 166

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 165
154 I N N O VAT I O N S I N T R AV E L D E M A N D M O D E L I N G , V O L U M E 2 focus group discussion best summarizes DRCOG's DRCOG VALIDATION AND SENSITIVITY philosophy: TESTING APPROACH Researchers and practitioners have not thought A plan describing the validation tests to be conducted for carefully enough about the criteria for validation the ABM components and the overall model system has of models. Researchers have the habit of asking been developed with the specification of the ABM for the practitioners to believe that activity-based methods region. It includes the standards by which the tests will will produce better impact assessment and fore- be evaluated, such as: casts because such models more appropriately rep- resent the actual decision process (we plead guilty Checks to ensure that the model component is pro- to this charge). There is a good basis for this line of ducing the correct results (i.e., verification of computa- thought, but researchers need to go beyond this tions); argument. They need to develop clear validation Comparisons of model parameters to comparable criteria and demonstrate the value of activity-based parameters in similar models in other areas; methods in ways that are easily understood. (2) Disaggregate validation of all model components estimated using disaggregate methods and comparing Because the ABM development process for the Den- the model outputs to the estimation data; ver region has just begun, this paper focuses on the ini- Testing of each model's sensitivity to variables tial plans for the validation and sensitivity testing of the through controlled modification of those input variables; models. Comparisons of the model component outputs to the results from the survey data set; and Comparisons, where data are available, of the base DRCOG ACTIVITY-BASED MODEL APPROACH year outputs from each model component to indepen- dent observed data (e.g., comparisons of mode choice The ABM approaches for the Denver region will be model outputs to linked trips estimated from transit based on those used in other parts of the country, partic- boarding counts). ularly the San Francisco Transportation Authority. Most of the components will be nested or multinomial logit The above tests are typical of model validation tests (MNL) models, sensitive to person and household demo- that have been recommended in documents such as the graphic variables and transportation level-of-service Model Validation and Reasonableness Checking Manual variables. Anticipated model components include: (1) and should be performed for all model development efforts. Synthetic population generator; Because there are more components in the proposed Regular workplace location choice model for each ABM than in a conventional trip-based model, there will worker; be significantly more component testing. It will be Regular school location choice model for each important to design validation tests that are appropriate student; for each component. While some tests will be analogous Household auto ownership choice model; to those performed for components of trip-based mod- Daily activity pattern choice model for each person- els, others will be different. Examples of similar tests day; include comparisons of modeled trip length frequencies Number of tours choice model for each person day; with those from the household survey (tour lengths must Work-based sub-tour generation; also be compared) and comparisons of modeled and Tour-level destination choice; observed mode shares. Examples of tests to be performed Tour-level mode choice; without comparable trip-based tests include the number Tour-level time-of-day choice; of trips per tour by purpose, amount of time spent in Trip-level destination choice; activities versus traveling on tours by purpose, and the Trip-level mode choice (conditional on tour mode number of activities performed by each person. The choice); and DRCOG model validation plan (3) provides a list of all Trip-level time-of-day choice (conditional on time tests to be performed. windows remaining after all previous choices). One difficulty in performing the tests is the lack of experience to determine standards. For example, how Several components will be transferred or adapted close should the modeled number of activities per person from the existing four-step model for the region. Exam- be to the observed number? In some cases, established ples include the area-type and parking cost models, and standards for trip-based models may be used to inform traffic and transit assignment procedures. the choice of standards for the ABM. In other cases, the