Access graph: a novel graph representation of public transport networks for accessibility analysis

Tina Šfiligoj*, Aljoša Peperko, Oded Cats · 2025

Read Access graph: a novel graph representation of public transport networks for accessibility analysis by Tina Šfiligoj*, Aljoša Peperko, Oded Cats (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-08361.txt.

Introduction

Tina Sfiligoj, Aljosa Peperko, and Oded Cats offer in 2025 a new way to draw a public transport network so that access is easy to read. Access, the ease of reaching places, drives transit use, yet measures of it lack a shared frame. Their access graph links two stops whenever the trip between them fits inside a set time budget.

In that drawing, a stop's number of links directly counts how many places a rider can reach in the budget. The team defines access and fairness scores from the graph and applies them to fifty-one metro networks worldwide. Watching how the graph fills as the budget grows reveals both network strength and rider experience.

Catalog record

CREATORTina Šfiligoj*, Aljoša Peperko, Oded Cats
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE41 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-08361

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Access graph: a novel graph representation of public transport networks for accessibility analysis Tina Šfiligoj*, Aljoša Peperko, Oded Cats Abstract Abstract Accessibility, defined as travel impedance between spatially dispersed opportunities for activity, is one of the main determinants of public transport (PT) use. In-depth understanding of its properties is crucial for optimal public transport systems planning and design. Although the concept has been around for decades and there is a large body of literature on accessibility operationalisation and measurement, a unified approach is lacking. To this end, we introduce a novel graph representation of public transport networks, termed the access graph, based on the shortest paths between nodes. Shortest paths are calculated using the in-vehicle time-weighted L- and frequency-weighted P-space representations to determine generalised travel times. Then there is an edge between two nodes in the access graph if the travel time between them is below a certain threshold time budget. In this representation, node degree directly measures the number of nodes reachable within a predetermined time. We study the threshold-dependent evolution of the access graph, focusing on average degree and degree distributions. Based on the topological properties of the access graph, we define a set of accessibility indicators. In addition, we propose indicators of access equity. We apply the methodology to a dataset of 51 metro networks worldwide. In all cases, a logistic-like growth of average degree with time budget is observed, indicating universal behaviour of accessibility and exhibiting the value of the proposed representation for unified accessibility studies and its potential for comparative analyses. We see a great potential for the access graph to drive in-depth studies of accessibility. Keywords : public transport, accessibility, network science, metro network, equity 1 Introduction Accessibility is key in providing individuals with the opportunity to take part in employment, education and recreational activities. The key objective of transport systems and public transport services in particular is therefore defined in terms of providing accessibility. The Hansen definition of accessibility as travel impedance between spatially dispersed opportunities for activity is at the core of defining accessibility indicators ( Hansen (1959) ; Geurs & Van Wee (2004) ) . It has been consistently in use since its conception as evident from a large body of literature, and has facilitated accessibility-based planning. Complementary to Hansen-like measures of access, also termed primal access, where travel time and number of reachable opportunities are the independent and dependent variable, respectively, dual access inverts the independent and dependent variables and measures the time needed to reach a predetermined number of nodes ( Cui & Levinson (2020) ) . Despite its wide theoretical and empirical developments, the exact role of accessibility remains somewhat unclear. While the general idea of accessibility is relatively straightforward, the concrete operationalisation and related indicators vary significantly across implementations. The exact definitions and functions of travel impedance differ, while determining reliable values for the number of opportunities is even more difficult to perform in a unified manner ( Lahoorpoor et al. (2022) ) . This has led researchers to call for unified approaches to accessibility. Levinson & Wu (2020) proposed a general model across different modalities and categories of activity. This was followed by proposing a method which allows for the comparison of different modes of urban accessibility across different cities ( Wu et al. (2021) ) . Graph-theoretical and network science approaches have been gaining in importance in public transport systems research. Although network science methods have been present in PTN research for almost four decades, having been introduced by Musso et al. (1988) ; Vuchic & Musso (1991) , and later advanced in a seminal paper by Von Ferber et al. (2009) , there has been a widely recognised concern that this research has been driven primarily by network scientists using PTNs as real-world examples for theoretical developments (e.g. ( Dupuy (2013) ) ). As a result, the practical implications for transportation researchers and planners have often been overlooked. Recently, this gap has begun closing down with transport researchers increasingly using network science approaches in modelling and studying PTN properties ( Derrible & Kennedy (2011) ; Ding et al. (2019) ) . This is especially evident in network robustness and vulnerability studies ( Adjetey-Bahun et al. (2016) ; Cats (2016) ; Cats et al. (2017) ; Cats & Krishnakumari (2020) ) . In contrast, network-based accessibility research has received comparatively less attention. Past studies focused either on analyzing the relative importance of nodes using centrality indicators ( Cats (2017) ; Šfiligoj et al. (2025) ) or compared different networks using an aggregate metric such as the average shortest path from each node to all other nodes, known as network (in)efficiency ( Dimitrov & Ceder (2016) ; de Regt et al. (2019) ) . In a pioneering study, Luo et al. (2019) suggested connecting network science and accessibility by calculating average shortest paths for each node to all other nodes, providing a Hansen-like indicator similar to closeness centrality. Their application of the method to eight tram networks worldwide demonstrated the potential of network-based methods for comparative assessment. As evident from recent literature, the increasing availability of standardised timetable data in the General Transit Feed Specification (GTFS) format has made accessibility studies increasingly accessible ( GoogleDevelopers (2025) ) . In the present work, we advance the argument for using graph-theoretical approaches for accessibility research. Accessibility is fundamentally about connectedness, which is naturally modelled with graphs. We argue that previous approaches that used aggregated values such as node centrality do not provide a sufficiently complete and insightful information when comparing accessibility across networks. First, centrality measures, even those in which the global properties of the network are implicit in their definition, offer a local and isolated node-level metrics. Second, centrality measures originate from social network analysis, and while they have been used successfully in PT studies, their interpretation in the transport context may be obscured to some extent. This may consequently limit the generalisation of these measures to potential new models, such as including land use, and thereby diminish their interpretability. To address the above-mentioned shortcomings, we propose a novel graph representation, called the access graph, where two nodes share an edge if they can be reached from one another within a given travel time budget. The matrix of generalised travel times is based on two standard PTN representations, the so-called L- and P-space representations, where nodes represent service stops. In L-space, or space-of-infrastructure, there is an edge between two nodes if they represent consecutive stops on a line, whereas in P-space, or space-of-service, there is an edge between each pair of nodes lying along the same line, reflecting the ability of passengers to travel between nodes without transfer. Edge weights may represent in-vehicle and waiting time, for L-space and P-space, respectively ( Luo et al. (2020) ) . The distance matrices of the corresponding representations are used to calculate generalised travel times. Importantly, in the resulting graph representation, while retaining the same set of nodes, the edges directly indicate accessibility between two nodes. Specifically, node degree in the access graph directly provides the number of reachable nodes within a specific time budget limit. Based on this representation, we study the temporal dependence of node degree distributions, focusing

Read the complete text