Date of Award

2017

Publication Type

Master Thesis

Degree Name

M.A.Sc.

Department

Civil and Environmental Engineering

Supervisor

Maoh, Hanna

Rights

info:eu-repo/semantics/openAccess

Abstract

Travel demand modeling is one of the key areas in transportation planning and engineering. Traditionally, it has been based on four inter-connected modules: trip generation, trip distribution mode choice, and traffic assignment. While the traditional approach remains popular among practitioners, it has been criticized widely due to its zonal aggregate nature. There has been a shift towards using micro-based models that use households and members of households as the units of analysis in lieu of the traffic analysis zones. This thesis contributes to advancing this micro-based paradigm by studying travel demand in the London Census Metropolitan Area (CMA), Ontario. It does so by developing an improved four-step travel demand model using a recent household travel survey that was collected in the year 2009. The focus will be as follows: first, compare various techniques that could be used to model trip generation (i.e., regression, cross-classification, discrete choice, and count models) at the micro-level. Also, compare the predictive ability of these micro-models against conventional zone-based models. Second, apply advanced geo-spatial methods and statistical techniques to model trip distribution using micro-data from the London Household Travel Survey (LHTS). To date, trip distribution in the four-stage model has relied on the gravity approach, which is too simplistic to capture the real complexities of spatial interactions between the traffic analysis zones forming an urban area. Also, its aggregate nature does not allow it to adequately capture the interaction of the traveler’s socio-economic characteristics with the attributes of alternative destinations. The results will allow us to devise an improved four-stage model that makes use of a conventional Household Travel Survey. Here, advanced techniques will be employed to improve the predictability of these models.

Share

COinS