2019 CSCE Annual Conference - Laval (Greater Montreal)

2019 CSCE Annual Conference - Laval (Greater Montreal) Conference

A Blending Approach for Trip Generation Rates Estimation: Minimum Sample Size for Special Generators

Mr. Feras Elsaid, Concordia University (Presenter)
Dr. Luis E Amador-Jimenez , Concordia University

Governments sometimes embark in expensive data collection campaigns to estimate local trip generation rates, given a suspicion that rates from well-established manuals are not applicable to local circumstances. This leads to inaccurate evaluation of the impacts that new developments introduce in the local transportation system. The estimation of trip rates requires collecting data from several sites with common characteristics; similar land use subcategories. However, certain land use subcategories may have limited number of local sites, such is the case with special generators. This paper presents an approach that blends rates from the few available local observations with rates from other databases or manuals. Moreover, this papers addresses the question of which manual to use to blend the rates and how many local observations are required to obtain an accurate estimation of local trip rate. An incremental approach, based on Full Bayesian, is followed to estimate the minimum required number of observations. Raw trip count data from previous studies are used to test the effect of blending rates from other manuals with rates estimated from local sites. The analysis in this study departs from the knowledge of the true rate, however, this rate is reserved for validation. The results show that at least three sites are required when the prior rate is borrowed from another manual of a region with similar characteristics and travel habits. However, up to 18 sites are required to obtain an acceptable estimate when the initial rate comes from a database of a dissimilar location.