PARAMETRIC ACCESS NETWORK MODEL

Parametric models are calibrated for the access portions of rail and bus trips. The models are designed to predict average zonal travel times as a function of the transportation system, zone size, and volume-related characteristics of a zone. The calibrated models are access walking, driving, and bus-riding time for rail trips and walking time to a stop for bus trips. Corresponding models are developed for the within-zone variance of the access time. These models provide input to the existing travel demand forecasting process by systematizing the way in which the access times are currently obtained for network coding. The importance of these values for travel forecasting has been repeatedly demonstrated in the past. These models also enable the use of large zones to help simplifying and speed up the transportation plan analysis and evaluation process. The predictive accuracy of the final models is evaluated in terms of standard indexes of forecasting accuracy. The results show that the coefficients of determination are high and that the coefficients of variation are low for all the models. Thus, the models should find an immediate use in transportation planning. /Author/