Orientations and matrix function-based centralities in multiplex network analysis of urban public transport

Kai Bergermann, Martin Stoll

Abstract

We study urban public transport systems by means of multiplex networks in which stops are represented as nodes and each line is represented by a layer. We determine and visualize public transport network orientations and compare them with street network orientations of the 36 largest German as well as 18 selected major European cities. We find that German urban public transport networks are mainly oriented in a direction close to the cardinal east-west axis, which usually coincides with one of two orthogonal preferential directions of the corresponding street network. While this behavior is present in only a subset of the considered European cities it remains true that none but one considered public transport network has a distinct north-south-like preferential orientation. Furthermore, we study the applicability of the class of matrix function-based centrality measures, which has recently been generalized from single-layer networks to layer-coupled multiplex networks, to our more general urban multiplex framework. Numerical experiments based on highly efficient and scalable methods from numerical linear algebra show promising results, which are in line with previous studies. The centrality measures allow detailed insights into geometrical properties of urban systems such as the spatial distribution of major transport axes, which can not be inferred from orientation plots. We comment on advantages over existing methodology, elaborate on the comparison of different measures and weight models, and present detailed hyper-parameter studies. All results are illustrated by demonstrative graphical representations.

Introduction

Understanding the growth and movement patterns within cities is a core challenge for urban planners and network scientists. While street networks are relatively straightforward to model as single-layer graphs, public transport networks (PTNs) are inherently more complex. A single physical stop often serves multiple transit lines, each operating with distinct frequencies and speeds. Traditional single-layer metrics often fail to capture this complexity, either by over-simplifying the connectivity or by failing mathematically on certain network structures.

This paper addresses these challenges by modeling transit systems as multiplex networks, where each transit line occupies its own layer. By applying matrix function-based centrality measures and scalable numerical linear algebra, the authors provide a robust framework for identifying critical infrastructure and analyzing the geometric orientations of urban transit.

Methodology

The research methodology integrates two primary geometric perspectives: orientation analysis and multiplex centrality modeling.

Overview of the methodology for analyzing urban public transport networks, from GTFS data processing to orientation and centrality visualization.

Network Orientations

To analyze the directional flow of transit, the researchers extracted stop sequences for every line to create directed graphs. By calculating compass bearings between consecutive stops, they generated "rose diagrams" to visualize preferential directions, allowing for a direct comparison with underlying street networks extracted from OpenStreetMap.

Multiplex Centrality Modeling

The PTN is represented as a multiplex network where nodes are physical stops and layers are individual transit lines.

  • Intra-layer edges connect stops along the same line, weighted by travel time and daily frequency.
  • Inter-layer edges represent transfers between different lines at the same physical location.

To identify important nodes, the authors utilized matrix exponentials and resolvents (such as Katz and Subgraph centrality). These measures count walks of all lengths while penalizing longer ones. To make these computations feasible for massive urban datasets, the study employs Krylov subspace methods (specifically the symmetric Lanczos method) and Gauss quadrature, ensuring the algorithms scale linearly with the number of node-layer pairs.

Key Results and Insights

The East-West Transit Bias

A striking finding of the study is the directional preference of European transit. While German street networks typically exhibit a cross-shape of two orthogonal preferential directions, their public transport networks overwhelmingly favor a single, blurred East-West axis. North-South preferential transit routes are surprisingly rare across the 54 major cities studied.

Rose diagrams showing the preferential orientations of public transport networks for 36 German cities, highlighting a dominant East-West bias.

Geography vs. Topology in Centrality

The analysis reveals that PTN centrality is "spiky" compared to the smooth radial decay often seen in street networks. Highly central stops cluster along major transit axes. A stop's importance is driven more by the daily frequency of the lines serving it than by the raw travel time between stops. Furthermore, the researchers found that introducing a realistic transfer penalty (e.g., 5 minutes) is essential; without it, centrality algorithms tend to over-concentrate importance on a single massive transfer hub.

Scalability and Tuning

The proposed Lanczos-based numerical methods achieved highly accurate centrality approximations in approximately 10 iterations, proving their viability for large-scale urban networks. Additionally, by tuning the hyper-parameters ($\alpha$ or $\beta$), the authors demonstrated the ability to shift the centrality map from a localized view (highlighting neighborhood hubs) to a global view (identifying critical city-wide infrastructure).

Conclusion

This work demonstrates that multiplex network analysis, combined with efficient numerical linear algebra, provides deeper insights into urban mobility than traditional methods. The discovery of a pervasive East-West bias in European transit orientations suggests significant differences between how cities are paved and how they are actually navigated. Future research may expand this framework to include directed edges for inbound/outbound analysis and incorporate multimodal layers, such as walking or car traffic, to provide a more holistic view of urban systems.

Summary updated

Abstract

We study urban public transport systems by means of multiplex networks in which stops are represented as nodes and each line is represented by a layer. We determine and visualize public transport network orientations and compare them with street network orientations of the $36$ largest German as well as $18$ selected major European cities. We find that German urban public transport networks are mainly oriented in a direction close to the cardinal east-west axis, which usually coincides with one of two orthogonal preferential directions of the corresponding street network. While this behavior is present in only a subset of the considered European cities it remains true that none but one considered public transport network has a distinct north-south-like preferential orientation. Furthermore, we study the applicability of the class of matrix function-based centrality measures, which has recently been generalized from single-layer networks to layer-coupled multiplex networks, to our more general urban multiplex framework. Numerical experiments based on highly efficient and scalable methods from numerical linear algebra show promising results, which are in line with previous studies. The centrality measures allow detailed insights into geometrical properties of urban systems such as the spatial distribution of major transport axes, which can not be inferred from orientation plots. We comment on advantages over existing methodology, elaborate on the comparison of different measures and weight models, and present detailed hyper-parameter studies. All results are illustrated by demonstrative graphical representations.
BibTeX
@article{bergermann2021orientations,
  author = {Kai Bergermann and Martin Stoll},
  title = {{Orientations and matrix function-based centralities in multiplex network analysis of urban public transport}},
  journal = {Appl. Netw. Sci.},
  year = {2021},
  doi = {10.1007/s41109-021-00429-9},
}

Loading related content...

Figure