Published in:Transportation Research Record: Journal of the Transportation Research Board
Language:
English
DOI:
10.3141/1993-10
ISSN:
0361-1981;
2169-4052
Origination:
Footnote:
Description:
<jats:p> Traditionally, metropolitan planning organizations are required to have their models calibrated on a continuing basis using new data. However, new survey data required to support these models do not exist in most urban areas. That makes it difficult to calibrate existing models or develop new travel demand models using emerging modeling techniques. As a result, the need to assess potential approaches and develop knowledge on how to transfer information collected in one context and use it in another context is becoming critical. This paper presents the process of developing a model that can facilitate household travel data transferability, which can reduce or eliminate the need for large data collection in the application context. Household records from the 2001 National Household Travel Survey are clustered into several homogeneous groups representing various types of households and their lifestyles. The clustering model is based on several variables that are easily obtainable from available sources. In addition to household characteristics, the variables of land use, built environment, and transportation system characteristics are included in the analysis. Various statistical measures of travel behavior are estimated for each cluster. By using an artificial neural network model, cluster membership rules and travel statistics are transferred to other geographic areas for the purpose of comparison and validation of the transferability method. </jats:p>