Spatial-Temporal Analysis and Forecasting of Urban Travel Demand

城市出行需求的時空分析及預測

Student thesis: Doctoral Thesis

View graph of relations

Author(s)

Detail(s)

Awarding Institution
Supervisors/Advisors
Award date12 Apr 2024

Abstract

Investigating urban travel demand holds significant importance owing to the pervasive traffic congestion in major cities, which is further compounded by the rapid pace of urbanization. Traffic congestion disrupts daily life, brings economic losses, and contributes to energy consumption and environmental pollution in cities, such as Shenzhen. Intelligent transportation systems (ITS) play a crucial role in improving safety and reducing congestion by integrating advanced technologies. Understanding urban travel demand and capturing accurate real-time data for traffic control are vital to the effective operations of the ITS. The effects of certain strategies on traveler behavior and network flow should be clarified to enhance ITS efficiency. Accurately forecasting travel demand is also critical for modern transportation systems due to the influence of ITS.

The urban travel demand of vehicles, including shared and single-occupancy vehicles, needs further investigation. Despite the recent advancements in understanding the spatial and temporal regularities and variations of urban travel demand, especially at aggregated levels, previous studies have mostly focused on taxi or online car-hailing services, such as Uber and DiDi, and neglected the broader spectrum of vehicle types, including shared and single-occupancy vehicles. Given the rising popularity of mobility as a service and its emphasis on shared and public transport, the travel demand patterns of different vehicle types need further examination.

This research investigates the impact of land use on the current and forecasted urban travel demand for shared and single-occupancy vehicles. Given the importance of land use in urban planning and transportation management, this study systematically analyzes its intrinsic relationship with urban travel demand to improve travel demand forecasting. By examining how land use influences urban transportation demand, this study contributes to the understanding of urban transportation system operations and provides a scientific foundation for urban planning and management. This research also explores the varying effects of different land use types on travel demand and examines ways to integrate these factors into forecasting models to improve their forecasting accuracy and support effective decision making in urban transportation management and planning.

First, this research analyzes the spatial and temporal aspects of urban travel demand patterns and uncover their key similarities. Spatial and temporal travel features are extracted from the origin–destination matrix using classical seasonal decomposition and singular value decomposition methods. This investigation deepens the current understanding of how urban travel demand evolves over time and space by identifying four primary patterns that illustrate these dynamics.

Second, this research investigates the relationships among different types of point of interest (POI) and urban travel demand and uses ordinary least squares and geographically weighted regression to understand their varying impacts across urban regions. The analysis identifies travel demand correlation POI linked to travel demand and reveals associations between POI and urban travel. These results also point to those key factors that influence urban traffic demand and provide actionable insights for urban traffic planning.

Lastly, this research thoroughly evaluates the performance of the attribute-enhanced spatiotemporal graph convolution model by using diverse data organization methods. Results show that varied data sources and detailed time granularity enhance the accuracy and applicability of travel demand forecasting, thus offering valuable insights for optimizing the performance of forecasting models.

This study also comprehensively examines the travel demand in Shenzhen, reveals its intricate spatial-temporal patterns, and offers crucial insights for understanding the dynamic changes in urban travel demand, optimizing urban transportation planning and management, and informing future research and practice.

    Research areas

  • Urban Travel Demand, Spatial-Temporal Analysis, POI, TDC-POI, Geographic Weighted Regression, Demand Forecasting, Data Organization