Do different datasets tell the same story about urban mobility — A comparative study of public transit and taxi usage

https://doi.org/10.1016/j.jtrangeo.2018.05.002Get rights and content

Abstract

Understanding human movements and their interactions with the built environment has long been a research interest in transport geography. In recent years, two important types of urban mobility datasets — smart card transactions and taxi GPS trajectories — have been used extensively but often separately to quantify travel patterns as well as urban spatial structures. Despite the fruitful research outcomes, the relationships between different types of transport flows in the same geographic area remain poorly understood. In this research, we propose an analytical framework to compare urban mobility patterns extracted from these two data sources. Using Singapore as a case study, this research introduces a three-fold comparative analysis to understand: (1) the spatial distributions of public transit and taxi usages and their relative balance; (2) the distance decay of travel distance, and (3) the spatial interaction communities extracted from the two transport modes. The research findings reveal that the spatial distributions of travel demand extracted from the two transport modes exhibit high correlations. However, more in-depth analysis (based on rank-size distribution and log odds ratio) reveals a higher degree of spatial heterogeneity in public transit usage. The travel distance of trips from public transit decays faster than that of taxi trips, highlighting the importance of taxis in facilitating long-distance travels. Both types of trips decay much faster when travel distance is beyond 20 km, which corresponds to the average distance from the urban periphery to the center. The spatial interaction communities derived from public transit are different on weekdays and weekends, while those of taxis show similar patterns. Both transport modes yield communities that reveal the city's polycentric structure, but their differences indicate that each of the transport modes plays a specific role in connecting certain places in the city. The study demonstrates the importance of comparative data analytics to urban and transportation research.

Introduction

The past two decades have witnessed an exponential growth of scientific research that characterizes human mobility and their interactions with the built environment. The rapid developments of information and pervasive sensing technologies have produced – especially in urban settings – a wide spectrum of human mobility datasets, empowering researchers to tackle critical questions in transport planning (Santi et al., 2014; Alexander et al., 2015; Tu et al., 2016), disease control (Bengtsson et al., 2011; Wesolowski et al., 2012), and social dynamics (Cho et al., 2011; Xu et al., 2017; Sun et al., 2013). The big data evolution has spurred “a new science” or many new sciences of cities, from which urban environments can be better understood as systems of networks and flows (Batty, 2013).

The networks and flows embedded in cities are defined by researchers through different types of datasets, resulting in a multi-faceted view of urban mobility patterns. For example, many studies have been conducted in recent years to quantify intra-urban mobility patterns based on taxicab usages (Wang et al., 2015; Liu et al., 2015; Kang and Qin, 2016), public transit data (Zhong et al., 2015; Liu et al., 2009), and mobile phone records (Gao et al., 2013; Ahas et al., 2010; Xu et al., 2016). Despite the fruitful research outcomes, most of the existing studies focus on a single type of human mobility dataset, which yields into insights that are somewhat isolated. The relationships between different types of networks and flows in the same geographic area – such as a city – remain poorly understood (Tu et al., 2018). It is, therefore, important to combine different data sources to obtain a comprehensive view of the spatial structures and organizations of cities. This will shed light on the bias when each data type is used alone to represent the dynamics of urban systems. More importantly, it would generate a deeper understanding of the interplay among different socio-economic processes.

In this research, we propose an analytic framework to compare urban travel patterns and the associated urban spatial structures extracted from smart card transactions and taxi GPS trajectories. The two types of datasets are widely used but often separately in revealing urban mobility patterns. Using Singapore as a case study, this work aims to fill the research gap by answering the following research question – do public transit and taxi usages in a city produce similar patterns of travel demand, travel distance, and urban spatial structures?

Based on origin-destination (OD) trips extracted from smart card transactions and taxi GPS trajectories — both cover a one-week period in Singapore — this study performs a three-fold comparative analysis. First, we analyze trip origins and destinations separately by focusing on the spatial distributions of outgoing and incoming trips. Two measures, namely rank-size distribution and log odds ratio, are introduced to quantify and compare spatial heterogeneity of travel demand extracted from the two datasets. We then examine spatial variations and statistical properties of travel distance (e.g., distance decay effect) to better understand the service radius of public transport and taxis in different parts of the city. Finally, we apply a community detection algorithm to the OD matrices to uncover the hidden spatial structures embedded in these two transport modes.

The remainder of this article is organized as follows. Section 3 provides an overview of related work of this research. Section 4 introduces the study area and the two mobility datasets. In Section 5, we introduce the approaches and measures for conducting the three-fold comparative analysis. We then present analysis results in Section 6. Finally, in Section 7, we conclude our findings and discuss future research directions.

Section snippets

Interplay between public transit and taxi services

Urban travel patterns are the outcome of the complex interactions between land use configuration and individual characteristics. The land use system governs spatial distribution of opportunities (in commercial, industrial areas) and the demand for these opportunities (in residential areas) (Geurs and van Wee, 2004). It determines intra-urban movement of people and goods from a macro level, which is modeled by trip generation and trip distribution in the “four-step” urban transportation planning

Related work

Understanding human mobility patterns has been a long standing research interest in areas such as urban planning, transportation, and geography. Before information and communication technologies (ICT) pervaded, travel surveys were used as the primary data source to support studies of human travel and daily activities. These studies cover important subjects such as trip chaining analysis (Hanson, 1980; Kitamura, 1984), characterization of human activity space (Newsome et al., 1998; Dijst, 1999;

Study area and datasets

Singapore is a city-state that covers a total area of 719 km2. It has a total population of 5.6 million as of 2016. The country has achieved rapid economic growth in the past century, and it is now a global finance and transport hub. The city is deployed with efficient mass transit services, allowing people to travel among destinations conveniently across the whole territory. According to the household interview travel survey (HITS) in 2012, mode share of public transit at peak period increased

Methods

Smart card transactions and taxi GPS trajectories capture different aspects of human mobility patterns. Both of them are well studied but separately in previous research. In this study, we propose an integrated framework to analyze and compare mobility patterns extracted from the two types of datasets. The comparison mainly focuses on the following perspectives: (1) the spatial distribution of travel demand; (2) the statistical properties of collective travel behavior (e.g., distance decay

Spatial distribution of travel demand

Fig. 2 illustrates the spatial distributions of average daily outgoing trips extracted from smart card transactions and taxi GPS trajectories on weekdays. From a visual perspective, it can be seen that the two types of travel demand match relatively well in geographic space. For example, subzones with a high level of public transit usage (Fig. 2A) tend to also produce a high number of taxi trips (Fig. 2B). This also happens in other combinations of trip category (incoming vs. outgoing) and day

Discussion and conclusion

Vast human mobility datasets have provided unprecedented opportunities for urban and transportation research. Such datasets have enabled the investigation of urban mobility patterns from a flow or network perspective. Existing studies attempt to reveal travel patterns and urban spatial structures from mobility data, but very few aim at comparing urban phenomena and processes from datasets of different types. In this research, we propose an analytical framework – by coupling smart card

Acknowledgement

The authors would like to thank the anonymous reviewers for their valuable comments and suggestions on earlier drafts of the manuscript. The research is supported in part by the National Research Foundation (NRF), Prime Minister's Office, Singapore, under CREATE programme, Singapore-MIT Alliance for Research and Technology (SMART) Centre, Future Urban Mobility (FM) Interdisciplinary Research Group, and the Hong Kong Polytechnic University Start-Up Grant under project 1-BE0J.

References (53)

  • Xi Liu et al.

    Revealing travel patterns and city structure with taxi trip data

    J. Transp. Geogr.

    (2015)
  • Marie-Pier Pelletier et al.

    Smart card data use in public transit: A literature review

    Transp. Res. C

    (2011)
  • Stefan Schönfelder et al.

    Activity spaces: measures of social exclusion?

    Transp. Policy

    (2003)
  • Wei Tu et al.

    Optimizing the locations of electric taxi charging stations: a spatial–temporal demand coverage approach

    Transport. Res. C

    (2016)
  • Wei Tu et al.

    Spatial variations in urban public ridership derived from gps trajectories and smart card data

    J. Transp. Geogr.

    (2018)
  • Wenjun Wang et al.

    A comparative analysis of intra-city human mobility by taxi

    Phys. A Stat. Mech. Appl.

    (2015)
  • Chen Zhong et al.

    Measuring variability of mobility patterns from multiday smart-card data

    J. Comput. Sci.

    (2015)
  • Michael Batty

    The New Science of Cities

    (2013)
  • Richard Becker et al.

    Human mobility characterization from cellular network data

    Commun. ACM

    (2013)
  • Linus Bengtsson et al.

    Johan Von Schreeb. Improved response to disasters and outbreaks by tracking population movements with mobile phone network data: a post-earthquake geospatial study in Haiti

    PLoS Med.

    (2011)
  • Vincent D. Blondel et al.

    Fast unfolding of communities in large networks

    J. Stat. Mech.

    (2008)
  • Vincent D. Blondel et al.

    A survey of results on mobile phone datasets analysis

    EPJ Data Science

    (2015)
  • Martijn Burger et al.

    Form follows function? Linking morphological and functional polycentricity

    Urban Stud.

    (2012)
  • Pablo Samuel Castro et al.

    Urban traffic modelling and prediction using large scale taxi gps traces

  • Eunjoon Cho, Seth A Myers, and Jure Leskovec. Friendship and mobility: user movement in location-based social networks....
  • Aaron Clauset et al.

    Finding community structure in very large networks

    Phys. Rev. E

    (2004)
  • Cited by (86)

    • Exploring the distance decay in port hinterlands under port regionalization using truck GPS data

      2024, Transportation Research Part E: Logistics and Transportation Review
    View all citing articles on Scopus
    View full text