More flexible models for measuring Travel Demand
rCITI Researcher: Dr Taha Hossein Rashidi
Modelling the demand for transport is a vital component of urban design and planning. But being able to model travel demand effectively and accurately is no easy task. There are multiple variables to consider, each which can have an effect on the others.
Ultimately the demand for transport comes about because of the activity happening at the end of the trip. The way modelling has conventionally occurred is through measuring the physical trips of people into transport analysis zones. This allows planners to develop mathematical models to forecast travel attributes based on all trips generated from and attracted to a zone.
As a major alternative to this traditional approach considers the activities and decision makers through another paradigm called “activity-based modelling” (based on discrete choice modelling). Over time, the mathematical complexity of the discrete choice models has grown rapidly resulting in equation-based models which are computationally intensive and do not necessarily reflect the behaviour of the decision maker. Thus, numerous simplifying assumptions are made to make these systems of models operational.
This research at rCITI explores several less computationally intensive methods that better reflect the behaviour of decision makers and can take into account the complexities resulted by interactions between agents. Basic concepts of learning-based methods and random graph models have only been employed in activity-based models in a limited manner while their usefulness has not been sufficiently examined.
Further, this research explores the practicality of complex adaptive system theory in travel demand modelling, as applications of this theory in fields other than transport are growing.
Major decisions modelled in an activity-based model along with methods that will be used to model them.