Interactions between Bus, Metro, and Taxi Use before and after the Chinese Spring Festival

Jianwei Huang, Xintao Liu, Pengxiang Zhao, Junwei Zhang, Mei-Po KwanISPRS Int. J. Geo Inf 2019 · DOI

TL;DR

This paper investigates how people's travel mode choices (bus, metro, taxi) change around the Chinese Spring Festival, a period of significant population movement. By analyzing smart card and GPS data in Shenzhen, they reveal dynamic interaction patterns between transit modes and how these patterns are influenced by the urban built environment.

What Problem Does This Paper Address?

Understanding how different public transit modes interact is crucial for designing sustainable urban transportation systems. While many studies focus on regular traffic conditions, the significant demographic shifts during events like the Chinese Spring Festival, where large numbers of people travel in and out of cities, can drastically alter these interactions. Existing research has largely overlooked how these transient demographic changes impact multi-modal transport patterns and their relationship with the urban built environment, especially using large-scale urban big data. This paper aims to fill this gap by examining these dynamics in Shenzhen, a city with a large migrant worker population.

Methods

The researchers employed a multi-step methodology:

  • Data Preprocessing: They cleaned and integrated smart card data (tap-in/tap-out records for bus/metro) and taxi GPS trajectory data. A key challenge was inferring bus passenger boarding points by cross-referencing bus line IDs, timestamps, and GPS data.
  • Interaction Pattern Identification: They used a spectral clustering algorithm to group traffic analysis zones (TAZs) based on the spatiotemporal similarity of transit mode usage patterns. This was done for both a "normal period" and the "Spring Festival period" to identify changes.
  • Elasticity Change Index (ECI): To quantify the change in transit mode patterns, they developed an ECI. This index measures how much a specific transit mode's ridership changes relative to the overall change in ridership across all modes, indicating sensitivity to demographic shifts.
  • Built Environment Analysis: They used Ordinary Least Squares (OLS) and Geographically Weighted Regression (GWR) models to explore the relationship between the calculated ECIs and various built environment characteristics (density, diversity, and design). GWR was particularly useful for capturing spatially varying relationships.

Datasets

The study utilized the following datasets for Shenzhen, China, around the 2017 Chinese Spring Festival (January 27 to February 11, 2017):

  • Smart Card Data (SCD): Tap-in/tap-out records for metro and bus services, including masked user IDs, timestamps, and station/line IDs.
  • Taxi GPS Trajectory Data: Including taxi ID, timestamp, latitude, longitude, and occupant status.
  • Traffic Analysis Zone (TAZ) Data: 491 TAZs covering the entire city, used as the primary spatial unit.
  • Road Network and Points of Interest (POI) Datasets: Obtained from OpenStreetMap (OSM) and Amap, used to characterize the built environment.

Key Insights and Ideas

  • Dynamic Transit Interactions: Transit mode interactions are not static; they change significantly in response to major demographic events like the Spring Festival.
  • Suburban vs. Central City Differences: Suburban areas with complex demographics and job opportunities for migrant workers showed higher sensitivity (ECI) to transit mode shifts, while the well-connected central city areas, heavily reliant on the metro, exhibited lower sensitivity.
  • Built Environment's Dual Role: The built environment influences both the baseline transit patterns and how these patterns change during special events. For instance, road density might stabilize patterns, while workplace density could lead to more dynamic shifts.
  • Mode-Specific Sensitivities: Bus and taxi ridership were found to be more sensitive to the influx/outflow of migrant workers compared to metro ridership.

Key Results

  • Pattern Shifts: Spectral clustering revealed distinct transit mode patterns for normal and Spring Festival periods. New patterns emerged, and others disappeared, indicating significant shifts in how people used different modes. For example, patterns P6 and P7 appeared during the Spring Festival, while P1 and P4 vanished.
  • ECI Distribution: The ECI analysis showed that bus and taxi ECIs were generally higher, particularly in suburban districts like Guangming, Longhua, Longgang, and Pingshan, suggesting greater adaptability or change in these modes during the festival.
  • Model Performance: GWR models significantly outperformed OLS models in explaining the relationship between ECIs and built environment variables, achieving adjusted R-squared values of 0.35 for taxi, 0.41 for bus, and a remarkable 0.93 for metro ECI.
  • Built Environment Impacts: Specific built environment factors had varied impacts. For example, road density and transit stop density were negatively associated with ECIs (promoting consistency), while workplace density was positively associated (promoting change). Residential density and distance to transit stops had different effects on taxi/bus versus metro ECIs.

Limitations and Future Work

  • Temporal Scope: The study's temporal coverage was limited to the Spring Festival and a normal period. Future work could explore other special periods with different cultural backgrounds (e.g., National Day).
  • Mode Inclusion: The analysis focused on bus, metro, and taxi. Future research could integrate a broader range of transport modes, including ride-sharing, walking, and private cars, for a more comprehensive understanding.
  • Cultural Context: The study's findings are specific to Shenzhen and its cultural context. Extending the analysis to other cities with different cultural backgrounds would be valuable.
  • Data Limitations: Integrating more detailed data on user demographics and trip purposes could further enhance the understanding of transit mode interactions.

Summary updated

BibTeX
@article{huang2019interactions,
  author = {Jianwei Huang and Xintao Liu and Pengxiang Zhao and Junwei Zhang and Mei-Po Kwan},
  title = {{Interactions between Bus, Metro, and Taxi Use before and after the Chinese Spring Festival}},
  journal = {ISPRS Int. J. Geo Inf.},
  year = {2019},
  doi = {10.3390/ijgi8100445},
}
Figure