Network-wide Spatio-Temporal Predictive Learning for the Intelligent Transportation System
- Publication Type:
- Thesis
- Issue Date:
- 2020
Open Access
Copyright Clearance Process
- Recently Added
- In Progress
- Open Access
This item is open access.
Large volumes of spatio-temporal data are increasingly collected and benefited to diverse domains, including transportation, urban optimization, community detection, climate science, etc. How to feed these large-scale data into a network-wide prediction model for the intelligent transportation system is a promising problem. Currently, even though a number of traffic prediction models have been proposed to enhance the travel services and improve operational performance of transit authorities, limited methods can be applied to forecast the network-wide traffic conditions afterward.
This thesis focuses on three problems in our predictive task. Firstly, the spatiotemporal data usually suffers from the missing data problem. Those missing values hide the useful information that may result in a distorted data analysis. In Chapter 3, a spatial missing data imputation method is proposed for multi-view urban statistical data. To address this problem, our method exploits an improved spatial multi-kernel clustering approach to guiding the imputation process cooperating with an adaptive-weight non-negative matrix factorization strategy. Secondly, in the crowd flow prediction, most existing techniques focus solely on forecasting entrance and exit flows of metro stations that do not provide enough useful knowledge for traffic management. In practical applications, managers desperately want to solve the problem of getting the potential passenger distributions to help authorities improve transport services, termed as crowd flow distribution (CFD) forecasts. Therefore, to improve the quality of transportation services, three spatiotemporal models are designed in Chapter 4 to effectively address the network-wide CFD prediction problem based on the online latent space (OLS) strategy. Our models take into account the various trending patterns and climate influences, as well as the inherent similarities among different stations that are able to predict both CFD and entrance and exit flows precisely. Lastly, with the development of urbanization, a real-world demand from transportation managers is to construct a new metro station in one city area that never planned before. Authorities are interested in the picture of the future volume of commuters before constructing a new station, and estimate how it would affect other areas. In this thesis, the specific problem is termed as potential passenger flow (PPF) prediction. Chapter 5 proposes a multi-view localized correlation learning method to provide a solution for the PPF prediction that can learn localized correlations via a multi-view learning process.
Please use this identifier to cite or link to this item: