
The Library
Road distance and travel time for spatial urban modelling
Tools
Crosby, Henry James (2018) Road distance and travel time for spatial urban modelling. PhD thesis, University of Warwick.
|
PDF
WRAP_Theses_Crosby_2018.pdf - Submitted Version - Requires a PDF viewer. Download (22Mb) | Preview |
Official URL: http://webcat.warwick.ac.uk/record=b3438515~S15
Abstract
Interactions within and between urban environments include the price of houses, the flow of traffic and the intensity of noise pollution, which can all be restricted by various physical, regulatory and customary barriers. Examples of such restrictions include buildings, one-way systems and pedestrian crossings. These constrictive features create challenges for predictive modelling in urban space, which are not fully captured when proximity-based models rely on the typically used Euclidean (straight line) distance metric.
Over the course of this thesis, I ask three key questions in an attempt to identify how to improve spatial models in restricted urban areas. These are: (1) which distance function best models real world spatial interactions in an urban setting? (2) when, if ever, are non-Euclidean distance functions valid for urban spatial models? and (3) what is the best way to estimate the generalisation performance of urban models utilising spatial data?
This thesis answers each of these questions through three contributions supporting the interdisciplinary domain of Urban Sciences. These contributions are: (1) the provision of an improved approximation of road distance and travel time networks to model urban spatial interactions; (2) the approximation of valid distance metrics from non-Euclidean inputs for improved spatial predictions and (3) the presentation of a road distance and travel time cross-validation metric to improve the estimation of urban model generalisation. Each of these contributions provide improvements against the current state-of-the-art. Throughout, all experiments utilise real world datasets in England and Wales, such datasets contain information on restricted roads, travel times, house sales and traffic counts. With these datasets, I display a number of case studies which show up to a 32% improved model accuracy against Euclidean distances and in some cases, a 90% improvement for the estimation of model generalisation performance.
Combined, the contributions improve the way that proximity-based urban models perform and also provides a more accurate estimate of generalisation performance for predictive models in urban space. The main implication of these contributions to Urban Science is the ability to better model the challenges within a city based on how they interact with themselves and each other using an improved function of urban mobility, compared with the current state-of-the-art. Such challenges may include selecting the optimal locations for emergency services, identifying the causes of traffic incidents or estimating the density of air pollution. Additionally, the key implication of this research on geostatistics is that it provides the motivation and means of undertaking non-Euclidean based research for non-urban applications, for example predicting with alternative, non-road based, mobility patterns such as migrating animals, rivers and coast lines. Finally, the implication of my research to the real estate industry is significant, in which one can now improve the accuracy of the industry's state-of-the-art nationwide house price predictor, whilst also being able to more appropriately present their accuracy estimates for robustness.
Item Type: | Thesis (PhD) | ||||
---|---|---|---|---|---|
Subjects: | H Social Sciences > HE Transportation and Communications Q Science > QA Mathematics T Technology > T Technology (General) |
||||
Library of Congress Subject Headings (LCSH): | Euclidean algorithm, Noise pollution, Travel time (Traffic engineering), City traffic | ||||
Official Date: | September 2018 | ||||
Dates: |
|
||||
Institution: | University of Warwick | ||||
Theses Department: | Department of Computer Science | ||||
Thesis Type: | PhD | ||||
Publication Status: | Unpublished | ||||
Supervisor(s)/Advisor: | Damoulas, Theodoros; Porto de Albuquerque, Joao; Jarvis, Stephen | ||||
Sponsors: | Engineering and Physical Sciences Research Council (EPSRC). Centre for Doctoral Training in Urban Science; Warwick Innovation Center. Assured Property Group | ||||
Format of File: | |||||
Extent: | xix, 147 leaves: illustrations, maps, charts | ||||
Language: | eng |
Request changes or add full text files to a record
Repository staff actions (login required)
![]() |
View Item |
Downloads
Downloads per month over past year