Simultaneous Travel Model Estimation from Survey Data and Traffic Counts

Abstract

This paper presents the successful application of a new method to improve travel demand forecasting models by taking advantage of cheap and readily available traffic count data and using them together with household travel survey data to inform the model's parameter estimates. Although traffic counts are frequently used in an ad hoc manner in the validation of travel model components, this paper presents a more rigorous, structured, and statistically efficient method to allow the information contained in traffic counts to influence the selection of model parameters simultaneously with household survey data. This formal process allows traffic counts to inform indirectly, but importantly, related parameters, such as destination choice utility functions, through formal statistical inference when human inference would be difficult, if not impossible, because of the complexity of the system and when manual random trial and error would be time- and cost-prohibitive. The approach used a genetic algorithm metaheuristic to implement a composite log likelihood and a pseudo composite log likelihood maximization in the development of a new travel model for the South Bend, Indiana, urban area for the Michiana Area Council of Governments. The process made use of a set of parameters transferred from another region and resulted in new parameters that produced significantly better consistency in the model with both local survey data and counts. Although computationally intense, this exciting new approach showed promise, at least for midsized urban areas.

Changes to the Transport Behaviour of Inhabitants of a Large City Due the Pandemic

“…Once transport investments have been made, such as changes to road infrastructure or the introduction of new transport services, both ITS data and survey results can be used to assess the effects of these interventions on population mobility, traffic congestion, travel times, and other indicators. Using ITS data and questionnaire survey results, it is possible to create forecasting models that help predict changes in population mobility behaviour in response to factors such as demographic changes, changes in transport policy, or the introduction of new transport technologies [72][73][74].…”

Individual Travel Knowledge Graph-Based Public Transport Commuter Identification: A Mixed Data Learning Approach

“…Neural network algorithms are among the most widely applied supervised learning methods in the field of machine learning and artificial intelligence. e method is well known in computer science, and there have been some successful applications of the method in traffic flow prediction [30], traffic model selection [31], and traffic congestion detection [32]. e BPNN method as a multilayer feedforward network is among the most widely applied supervised classification methods.…”

Traffic Flow Analysis and Solutions to Ease Traffic Flow at Unsignalized Taxila Intersection

“…Simultaneous Travel Model was estimated from survey data and traffic counts. They presented the successful application of a new method to improve travel demand forecasting models by taking advantage of cheap and readily available traffic count data and using them together with household travel survey datato inform the model's parameter estimates (Bernardin et al, 2012).…”