Where are public transit needed - Examining potential demand for public transit for commuting trips

This study investigates the potential demand for public transit for commuting trips – a rarely explored perspective for the study of public transit systems. Potential demand for public transit refers to the kind of demand that is not explicitly expressed or realized but will be present if condition permits (i.e. when public transit facility is accessible). In this study, potential demand is surrogated by the proportion of people who may potentially use public transit as primary transportation mode, once public the transit facility becomes available and accessible to these people. A Need Index is introduced to measure the relative magnitude of potential demand. Multiple regression is employed to identify predictive variables for the share of public transportation for work trips. Given the predictive variables, two independent methods are developed to examine potential demands across the space. One method is the abovementioned Need Index method and the other is a data mining approach. The Need Index is mathematically modeled which computes a numeric measure for each spatial unit. The second method uses self-organizing maps (SOM) to find clusters in the high-dimensional vector space of the predictive variables. The study then presents a cross-examination analysis between the results from two methods. An empirical study of Atlanta, Georgia is carried out. In the case study, The United States Census Transportation Planning Package 2000 data at the traffic analysis zones (TAZ) level are used. Existing transit network data are prepared and preprocessed in GIS for spatial analysis. The results of both methods are compared and displayed in GIS for visual examination of spatial distribution of the potential demands. A critique on the comparison of advantages and shortcomings between both methods is provided. It shows that the need index is superior due to its simplicity and its higher level of measurement of potential demands. 2007 Elsevier Ltd. All rights reserved. 0198-9715/$ see front matter 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.compenvurbsys.2007.08.005 * Fax: +1 706 542 2388. E-mail address: [email protected] 536 X. Yao / Comput., Environ. and Urban Systems 31 (2007) 535–550