Examining Day-to-Day Dynamic Transit Accessibility Using Functional Data Analysis

Abstract

Measurements of transit accessibility have been used to evaluate the interactions between transit networks and land use and, more recently, to identify marginalized neighborhoods in need of transportation services to support urban policymaking, transportation planning, and the evaluation of transit systems. Because transit systems are dynamic and their weekly use patterns have high temporal variation, their accessibility is also highly variable across days. In this study, we propose a set of dynamic transit accessibility measures that examine the temporal fluctuations of accessibility by public transit modes. To this end, we employ a novel approach of the functional data analysis that takes cross-area time-dependent curves fitted from the daily accessibility values. We examine Seoul’s intermodal integrated transit system to measure dynamic transit accessibilities to capture the heterogeneity of day-to-day fluctuations in accessibility over the course of a week across neighborhoods. The spatial representation of the dynamic transit accessibilities and the clusters based on the measures reveal an urban structure on which neighborhoods are differentiated by temporal fluctuations of accessibility, not just by a traditional concentric urban pattern of accessibility.

公交可达性指数被用于评估公交网络与土地利用之间的相互作用, 近来还用于确定需要交通服务的边缘化社区, 支持城市决策、交通规划和公交系统评估。由于公交系统是动态的、每周的公交使用模式在时间上具有高度差异性, 因此公交可达性也有很大的日变化。本研究提出描述公交可达性在时间上波动的一组动态公交可达性指数。为此, 我们利用日可达性数据拟合的跨区域时间曲线, 采用了新的函数型数据分析方法（Functional Data Analysis, FDA）。本文研究韩国首尔的跨模式综合交通系统, 度量了动态交通可达性, 获取了一周内跨社区可达性日波动的差异。动态公交可达性的空间展现和基于这些度量的集群, 揭示了这样的城市结构：基于可达性时间波动（而不是传统同心圆可达性城市模式）的社区划分。

Las mediciones de accesibilidad del tránsito han sido usadas para evaluar las interacciones entre las redes de tránsito y el uso del suelo, y, más recientemente, para identificar los vecindarios marginales carentes de servicio de transporte, con el fin de respaldar la formulación de políticas urbanas, la planificación del transporte y la evaluación de los sistemas de tránsito. Por cuanto los sistemas de tránsito son dinámicos y debido a que sus patrones de uso semanal tienen alta variación temporal, su accesibilidad es también muy variable a lo largo de los días. En este estudio, proponemos un conjunto de medidas dinámicas de accesibilidad del tránsito, que examinan las fluctuaciones temporales de la accesibilidad según los modos de transporte público. Con esa finalidad, empleamos un enfoque novedoso del análisis de datos funcionales que toma curvas transversales dependientes del tiempo ajustadas a partir de los valores de accesibilidad diarios. Examinamos el sistema de tránsito intermodal integrado de Seúl, para medir las accesibilidades dinámicas de tránsito, con la mira de captar la heterogeneidad de las fluctuaciones diarias de accesibilidad en el curso de una semana, a través de los vecindarios. La representación espacial de las accesibilidades dinámicas del tránsito, y las agrupaciones basadas en las medidas, revelan una estructura urbana donde los vecindarios se diferencian por las fluctuaciones temporales de la accesibilidad y no solo por un patrón urbano concéntrico tradicional de accesibilidad.

Key Words:

• dynamic transit accessibility
• dynamic transit cluster
• functional data analysis
• public transit
• Seoul

关键词:

• 动态公交可达性
• 动态公交集群
• 函数型数据分析
• 公交
• 首尔。

Palabras clave:

• accesibilidad dinámica del tránsito
• análisis de datos funcionales
• conglomerado dinámico de tránsito
• tránsito público

Notes

1 In 2015, the total length of roads in Seoul was 8,215 km (Seoul Metropolitan Government 2000), and the number of bus stops and subway stations were 13,518 and 364, respectively. This indicates transit stops exists on roads every 592 m, enabling the residents to conveniently access the transit system.

2 The study period was intentionally selected to avoid major holidays and extreme weather conditions that could affect the accessibility pattern. The original smart card data can be obtained from the Seoul Metropolitan Government. The smart card records forty-seven properties, such as riding time and disembarkation time at station, payment information at each time, and transfer counts (M. Kim and Lee 2017). With that information, a trajectory of each journey was constructed as a graph, and its travel time from the beginning to the end was calculated based on the algorithm by Lee et al. (2019) and Park and Lee (2015).

Additional information

Funding

This work was supported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2018S1A5A8026828).

Notes on contributors

Paul H. Jung

PAUL H. JUNG is a PhD Candidate in the Department of Geography and Earth Sciences at the University of North Carolina at Charlotte, Charlotte, NC 28223. E-mail: [email protected]. His research interests include international trade, transportation, neighborhood dynamics, spatial statistics, and spatial data uncertainty.

Hyun Kim

HYUN KIM is an Associate Professor in the Department of Geography at the University of Tennessee, Knoxville, TN 37996. E-mail: [email protected]. His research interests include transportation and location modeling using spatial optimization and GIScience.

Keumsook Lee

KEUMSOOK LEE is an Emeritus Professor in the Department of Geography at Sungshin Women’s University, Seoul, Korea 02844. E-mail: [email protected]. Her research interests include the dynamic complexities in transportation systems and land-use patterns and the issue of online influences on the offline space.

Yena Song

YENA SONG is an Associate Professor in the Department of Geography at Chonnam National University, Gwangju 61186, Republic of Korea. E-mail: [email protected]. Her research centers around the public transportation, travel behavior, accessibility, and real estates.