pith. machine review for the scientific record. sign in

arxiv: 2604.15784 · v2 · submitted 2026-04-17 · ⚛️ physics.soc-ph

Recognition: no theorem link

Comparing Analytical Approaches for Bike Station Expansion: A Location-Allocation Study in Trondheim, Norway

M. Tsaqif Wismadi , Oluwaleke Yusuf , Adil Rasheed , Yngve Karl Fr{\o}yen , Thomas Alexander Sick Nielsen
Authors on Pith no claims yet

Pith reviewed 2026-05-12 02:53 UTC · model grok-4.3

classification ⚛️ physics.soc-ph
keywords bike-sharinglocation-allocationspatial decision supporturban mobilityTrondheimoptimizationsuitability analysisnetwork planning
0
0 comments X

The pith

Methodological choices in bike station models lead to different spatial priorities and expansion recommendations in Trondheim.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The study compares three approaches to placing bike-sharing stations: weighted linear combination, the maximal covering location problem, and a data-driven suitability score. All three use the same set of 24 spatial features and a shared weighting scheme to design a new 68-station network from scratch. The resulting maps show clear differences, with one emphasizing population and transit coverage, another spreading stations for broad reach, and the third balancing demand and access. These model outputs differ from the actual existing network in the city. A combined view identifies 12 consensus sites for future expansion that feature good connections and unmet demand.

Core claim

When the weighted linear combination, maximal covering location problem, and exogenous spatial feature suitability score are each applied to the same 24 features to site 68 stations, they generate systematically different networks. The weighted approach covers demand areas most effectively, the covering problem maximizes geographic distribution, and the suitability score integrates intensity with accessibility. None matches the current bike network closely, indicating that non-analytical factors have shaped existing placements. Synthesis of the three identifies twelve priority expansion locations based on shared strengths in multimodal access and underserved areas.

What carries the argument

The unified framework that runs the three models in parallel on identical inputs to isolate the effect of decision logic.

If this is right

  • Planners using different models will recommend different sets of station locations even with the same data.
  • The current network in Trondheim leaves room for improvement according to all three analytical methods.
  • Consensus sites combine strengths from multiple approaches and may offer more robust choices.
  • Method selection affects how well the network serves population centers versus remote areas.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Testing these methods in other cities could reveal whether the observed differences are consistent across urban contexts.
  • Adding real-time usage data or citizen input might help reconcile the varying priorities from the models.
  • Planners could run sensitivity tests on the feature weights to see how much the outputs change.

Load-bearing premise

The assumption that the selected 24 spatial features together with the fixed hierarchical weights fully and fairly represent all factors that determine good bike station locations.

What would settle it

Collecting actual ridership or demand data at the 12 consensus sites and at sites favored by only one model to check if the consensus locations indeed show higher usage potential.

Figures

Figures reproduced from arXiv: 2604.15784 by Adil Rasheed, M. Tsaqif Wismadi, Oluwaleke Yusuf, Thomas Alexander Sick Nielsen, Yngve Karl Fr{\o}yen.

Figure 1
Figure 1. Figure 1: Four-stage methodological framework adopted in the study. 2.1. Study Area Overview Trondheim constitutes a demanding yet informative setting for evaluating alternative analytical approaches to bike￾sharing station placement. As a mid-sized municipality in central Trøndelag county, the city covers approximately 496.9 square kilometres and had an estimated population of 216,518 inhabitants in 2024, yielding … view at source ↗
Figure 2
Figure 2. Figure 2: (a) Distribution of Trondheim station locations; and (b) Current station density and clustering. The bike-sharing scheme operates alongside a wider shared micromobility market. In addition to public bicycles, three private e-scooter operators provide services across Trondheim, with substantial spatial overlap and similar trip purposes, particularly within central areas. However, e-scooter use is constraine… view at source ↗
Figure 3
Figure 3. Figure 3: Spatial grid framework adopted for feature engineering and location-allocation modelling. Each cell is represented by its centroid, and all spatial features are aggregated or calculated at the cell level. Through this spatial feature engineering process, 24 explanatory variables were derived and organised into seven conceptual dimensions that represent different mechanisms influencing cycling demand and bi… view at source ↗
Figure 4
Figure 4. Figure 4: Two-stage SSE modelling workflow in which exogenous features are used to predict mobility indicators which are aggregated into a composite suitability score for station placement. All flow targets are log-transformed prior to training, 𝑦 ′ 𝑖 = log(1 + 𝑦𝑖 ), to stabilise variance across the distribution and reduce the influence of high-flow outliers during optimisation. The inverse transformation is applied… view at source ↗
Figure 5
Figure 5. Figure 5: (a) WLC proposed station locations; and (b) WLC station kernel density and clustering. M.T. Wismadi et al.: Preprint submitted to Elsevier Page 12 of 28 [PITH_FULL_IMAGE:figures/full_fig_p012_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: (a) MCLP proposed station locations; and (b) MCLP station kernel density and clustering. 3.3. SSE Proposed Locations At first glance, the SSE placement results appear to represent a hybrid configuration between the WLC and MCLP approaches. With respect to its orientation towards the city centre, SSE resembles the pattern produced by WLC. The model maintains a clear central emphasis, with its main cluster c… view at source ↗
Figure 7
Figure 7. Figure 7: (a) SSE proposed station locations; and (b) SSE station kernel density and clustering. 3.4. Model Comparison The modelling approaches not only produced alternative station configurations and spatial distributions, but also generated a suitability score for each selected location. These scores represent the relative desirability of locations under the logic of each model and therefore provide a common basis… view at source ↗
Figure 8
Figure 8. Figure 8: Spearman correlation (𝜌) between spatial features and suitability scores for WLC, MCLP, and SSE selected locations (|𝜌| = 1 indicates strongest association). models, indicating that locations with higher existing cycling activity are strongly prioritised. Transit-related indicators also display high correlations, especially in SSE, where boarding, alighting, and aggregated transit flow contribute strongly … view at source ↗
Figure 9
Figure 9. Figure 9: Relative ranking of feature–score association strength across WLC, MCLP, and SSE models, based on absolute Spearman correlation with suitability scores (rank 1 = strongest association) [PITH_FULL_IMAGE:figures/full_fig_p016_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Proposed future station locations based on consensus clustering of WLC, MCLP, and SSE selected sites. The map shows the 12 identified clusters, with cluster medoids representing the approximate new station locations. M.T. Wismadi et al.: Preprint submitted to Elsevier Page 18 of 28 [PITH_FULL_IMAGE:figures/full_fig_p018_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: presents the underlying spatial feature values for each consensus zone alongside the existing station network, complementing the overview in [PITH_FULL_IMAGE:figures/full_fig_p019_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Relative service performance ranking of existing and proposed station networks across spatial features (wider radial level indicates a better rank). The comparison results suggest that the WLC configuration provides the strongest candidate for maximising aggregated service exposure under the assumed walking threshold, particularly when the planning objective is to maximise interaction with population, lan… view at source ↗
read the original abstract

The strategic placement of bike-sharing infrastructure shapes urban accessibility and mobility outcomes. However, station-allocation approaches vary in their assumptions and decision logic. This study examines how alternative modelling paradigms prioritise urban space when applied to the same planning problem in Trondheim, Norway. We developed a unified analytical framework to compare three location-allocation approaches: weighted linear combination (WLC), maximal covering location problem (MCLP), and a data-driven suitability score based on exogenous spatial features (SSE). Each model designs a 68-station bike-sharing network from scratch using the same 24 spatial features and hierarchical weighting scheme. The resulting configurations are compared with the existing network, and consensus-based synthesis identifies 12 priority locations for expansion. The findings reveal systematic differences in spatial prioritisation across modelling approaches. WLC achieves the strongest coverage of population and transit demand, MCLP produces the widest spatial distribution prioritising geographic reach, and SSE balances demand intensity with accessibility. All model-derived configurations diverge from the existing network, highlighting the influence of historical and institutional factors on real-world deployment. Consensus synthesis identifies 12 expansion sites characterised by multimodal integration potential, underserved residential clusters, and high latent demand. This analysis demonstrates that methodological choices fundamentally shape spatial decision-support outcomes. By systematically evaluating classical optimisation and data-driven approaches under controlled conditions, the study provides evidence-based recommendations for bike-sharing network expansion and clarifies the strengths and limitations of alternative analytical frameworks for location-allocation planning.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper claims that applying three location-allocation approaches—weighted linear combination (WLC), maximal covering location problem (MCLP), and a data-driven suitability score (SSE)—to identical inputs of 68 stations, 24 spatial features, and a hierarchical weighting scheme for bike-sharing network design in Trondheim produces systematically different spatial configurations. WLC prioritizes population and transit coverage, MCLP maximizes geographic spread, SSE balances intensity and access; all diverge from the existing network, and a consensus synthesis identifies 12 priority expansion sites, demonstrating that methodological choices fundamentally shape planning outcomes.

Significance. If the results hold, the work is significant for providing a controlled, apples-to-apples comparison of classical optimization and data-driven paradigms under fixed inputs, which directly isolates the effect of model assumptions on spatial recommendations. This offers practical, evidence-based guidance for bike-sharing expansion while clarifying the distinct strengths of each approach. The unified framework and explicit divergence from the real network are notable strengths.

major comments (2)
  1. [Abstract] Abstract: The abstract (and by extension the methods) provides no details on the data sources for the 24 spatial features, any validation against observed usage or ridership data, sensitivity tests on the hierarchical weighting scheme, or the exact procedure for the consensus-based synthesis that yields the 12 priority locations. These omissions are load-bearing for the strength of the evidence-based recommendations.
  2. [Methods] Methods: The hierarchical weighting scheme for the 24 features is treated as fixed without reported sensitivity analysis; a test varying the weights within plausible ranges would be needed to confirm that the observed differences in spatial configurations (WLC vs. MCLP vs. SSE) are robust rather than artifacts of the particular weighting.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and positive assessment of the manuscript's significance. We address each major comment below and have made revisions to improve transparency and robustness.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The abstract (and by extension the methods) provides no details on the data sources for the 24 spatial features, any validation against observed usage or ridership data, sensitivity tests on the hierarchical weighting scheme, or the exact procedure for the consensus-based synthesis that yields the 12 priority locations. These omissions are load-bearing for the strength of the evidence-based recommendations.

    Authors: We appreciate the referee drawing attention to these elements. Data sources for the 24 spatial features are specified in Section 3.1 of the manuscript (Statistics Norway, OpenStreetMap, Trondheim municipal GIS layers, and national transport registries). The study is a controlled comparison of planning models rather than a validation exercise against ridership data; we have added an explicit statement to this effect in the revised discussion section. The consensus synthesis procedure is detailed in Section 4.4 as an overlay of the top 20 sites from each model, with the 12 priority locations defined by sites appearing in at least two models. We have expanded the abstract to include concise references to data provenance and the consensus method. revision: yes

  2. Referee: [Methods] Methods: The hierarchical weighting scheme for the 24 features is treated as fixed without reported sensitivity analysis; a test varying the weights within plausible ranges would be needed to confirm that the observed differences in spatial configurations (WLC vs. MCLP vs. SSE) are robust rather than artifacts of the particular weighting.

    Authors: We agree that explicit sensitivity testing strengthens the claims. In the revised manuscript we have added a new subsection (3.3.2) describing the hierarchical weighting derivation via analytic hierarchy process with local planners. We also performed a limited sensitivity analysis by varying the weights of the four highest-ranked features (population, transit proximity, residential density, and commercial land use) by ±15% and ±30% and re-running the three models. Results are reported in new Appendix C; the systematic differences between WLC, MCLP, and SSE persist, and 10 of the 12 consensus sites remain stable. This supports the robustness of the core findings while acknowledging that exhaustive weight perturbation across all 24 features was beyond the scope of the current study. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper applies three independent location-allocation models (WLC, MCLP, SSE) to identical external inputs consisting of 68 stations, the same 24 spatial features, and a fixed hierarchical weighting scheme. Each model is run separately on these inputs to generate distinct spatial configurations, which are then compared against each other and the existing network. No derivation step reduces by construction to a self-definition, a fitted parameter renamed as a prediction, or a self-citation chain; the differentiated outcomes follow directly from the distinct optimization logics applied to the same external data. The central claim that methodological choices shape outcomes is therefore supported by the controlled comparison without internal circularity.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on standard domain assumptions about spatial data sufficiency rather than new inventions or heavy fitting. The hierarchical weighting scheme introduces one adjustable element but is applied uniformly.

free parameters (1)
  • hierarchical weighting scheme for 24 features
    Weights are chosen hierarchically to combine the spatial features but specific values or derivation details are not provided.
axioms (1)
  • domain assumption The 24 spatial features sufficiently represent demand, accessibility, and suitability factors for bike stations
    This assumption is invoked to justify using the same inputs across all three models.

pith-pipeline@v0.9.0 · 5588 in / 1323 out tokens · 85312 ms · 2026-05-12T02:53:58.469031+00:00 · methodology

Review history (2 revisions) →

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 28 canonical work pages

  1. [1]

    URL:https://www.atb.no/rapporter/na- planlegges-fremtiden-ny-rutestruktur/

    Future Route Structure including Superbus for Greater Trondheim 2019-2029. URL:https://www.atb.no/rapporter/na- planlegges-fremtiden-ny-rutestruktur/. AtB AS,

    work page 2019

  2. [2]

    URL:https://www.atb.no/trondheimsomradet/

    Trondheim Area Timetables. URL:https://www.atb.no/trondheimsomradet/. Aydın,Z.E.,Erdem,B.İ.,Erzurumcıcıek,Z.I.,2023. Predictionbike-sharingdemandwithgradientboostingmethods. PamukkaleUniversityJournal of Engineering Sciences 29, 824–832. doi:10.5505/pajes.2023.39959. Bahadori, M., et al.,

    work page doi:10.5505/pajes.2023.39959 2023

  3. [3]

    ISPRS International Journal of Geo- Information doi:10.3390/ijgi10080554

    A systematic review of station location techniques for bike-sharing systems. ISPRS International Journal of Geo- Information doi:10.3390/ijgi10080554. Banerjee, S., et al.,

    work page doi:10.3390/ijgi10080554

  4. [4]

    Transport Findings doi:10.32866/ 001c.13155

    Optimal locations for bikeshare stations: A gis-based location-allocation approach. Transport Findings doi:10.32866/ 001c.13155. Cai,X.,etal.,2025. Differencesinbike-sharingusageanditsassociationswiththebuiltenvironment. TransportPolicydoi: 10.1016/j.tranpol. 2025.01.004. Carlsson, J., Jia, F.,

    work page doi:10.1016/j.tranpol 2025

  5. [5]

    European Journal of Operational Research 256, 615–630

    A review of p-median and related facility location problems. European Journal of Operational Research 256, 615–630. doi:10.1016/j.ejor.2016.06.023. Chen, T., Guestrin, C.,

    work page doi:10.1016/j.ejor.2016.06.023 2016

  6. [6]

    XGBoost: A scalable tree boosting system, in: Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, ACM, New York, NY, USA. pp. 785–794. doi:10.1145/2939672.2939785. Church, R., ReVelle, C.,

    work page doi:10.1145/2939672.2939785

  7. [7]

    Papers in Regional Science 32, 101–118

    The maximal covering location problem. Papers in Regional Science 32, 101–118. doi:10.1111/j.1435-5597. 1974.tb00902.x. Dehdari Ebrahimi, Z., et al.,

    work page doi:10.1111/j.1435-5597 1974

  8. [8]

    Transport Policy doi:10.1016/j.tranpol.2022.04.002

    Using a gis-based spatial approach to determine optimal bike-share station locations. Transport Policy doi:10.1016/j.tranpol.2022.04.002. Duran-Rodas, D., et al.,

    work page doi:10.1016/j.tranpol.2022.04.002 2022

  9. [9]

    Computers & Industrial Engineering 62, 368–407

    Covering problems in facility location: A review. Computers & Industrial Engineering 62, 368–407. doi:10.1016/j.cie.2011.08.020. Grau-Escolano, J., et al.,

    work page doi:10.1016/j.cie.2011.08.020 2011

  10. [10]

    Operations Research 12, 450–459

    Optimum locations of switching centers and the absolute centers and medians of a graph. Operations Research 12, 450–459. doi:10.1287/opre.12.3.450. Hampshire, R., et al.,

    work page doi:10.1287/opre.12.3.450

  11. [11]

    Transportation Research Record doi:10.3141/2247-07

    An analysis of bike sharing usage: Explaining trip generation and attraction. Transportation Research Record doi:10.3141/2247-07. Hauke, J., Kossowski, T.,

    work page doi:10.3141/2247-07

  12. [12]

    Quaestiones Geographicae 30, 87–93

    Comparison of values of pearson’s and spearman’s correlation coefficients on the same sets of data. Quaestiones Geographicae 30, 87–93. doi:10.2478/v10117-011-0021-1. Karatas, M., Razi, N., Tozan, H.,

    work page doi:10.2478/v10117-011-0021-1

  13. [13]

    Technical Gazette 24, 399–407

    A multi-criteria assessment of the p-median, maximal coverage and p-center location models. Technical Gazette 24, 399–407. doi:10.17559/TV-20160508131905. Karteek,G.,Schwai,M.,2022. Aredenseresidentialneighborhoodsliveable?AcomparativeanalysisofoldandnewneighborhoodsinTrondheim. Journal of Architecture and Civil Engineering 7, 124–138. URL:https://www.que...

    work page doi:10.17559/tv-20160508131905 2022

  14. [14]

    Transport Reviews doi:10.1080/01441647.2023

    The synergy of bicycles and public transport: A systematic review. Transport Reviews doi:10.1080/01441647.2023. 2222911. Li, Y., et al.,

    work page doi:10.1080/01441647.2023 2023

  15. [15]

    Land doi:10.3390/land14061291

    Exploring the relationship between the built environment and bike-sharing usage. Land doi:10.3390/land14061291. Malczewski, J.,

    work page doi:10.3390/land14061291

  16. [16]

    John Wiley & Sons, New York

    GIS and Multicriteria Decision Analysis. John Wiley & Sons, New York. doi:10.1111/j.1538-4632.2002.tb01077.x. Malczewski,J.,2006. Gis-basedmulticriteriadecisionanalysis:asurveyoftheliterature. InternationalJournalofGeographicalInformationScience 20, 703–726. doi:10.1080/13658810600661508. Meiforth, J.J.,

    work page doi:10.1111/j.1538-4632.2002.tb01077.x 2002

  17. [17]

    Miljøpakken

    Technical Report. Miljøpakken. URL:https://trondheimsregionen.no/wp-content/ uploads/2023/10/rvu-trondheimsregionen-2022.pdf. Minieka, E.,

    work page 2023

  18. [18]

    SIAM Review 12, 138–139

    The m-center problem. SIAM Review 12, 138–139. doi:10.1137/1012016. Nasri, A., Zhang, L.,

    work page doi:10.1137/1012016

  19. [19]

    Journal of Transport and Land Use doi:10.5198/jtlu.2020.1615

    Analysis of the effect of multi-level urban form on bikeshare demand. Journal of Transport and Land Use doi:10.5198/jtlu.2020.1615. NGI,

    work page doi:10.5198/jtlu.2020.1615 2020

  20. [20]

    Transport Geography doi:10.1016/j.jtrangeo.2020.102870

    Bikeshare destination choices and accessibility. Transport Geography doi:10.1016/j.jtrangeo.2020.102870. Rixey, R.,

    work page doi:10.1016/j.jtrangeo.2020.102870 2020

  21. [21]

    Transportation Research Record doi:10.3141/2387-09

    Station-level forecasting of bikesharing ridership. Transportation Research Record doi:10.3141/2387-09. Sentralbyrå, S.,

    work page doi:10.3141/2387-09

  22. [22]

    Technical Report 2024:9

    Towards a Grid-Based Nordic Territorial Typology. Technical Report 2024:9. Nordregio. doi:10.6027/R2024:91403-2503. Thon Group,

    work page doi:10.6027/r2024:91403-2503 2024

  23. [23]

    Operations Research 19, 1363–1373

    The location of emergency service facilities. Operations Research 19, 1363–1373. doi:10.1287/opre.19.6.1363. M.T. Wismadi et al.:Preprint submitted to Elsevier Page 27 of 28 Comparing Analytical Approaches for Bike Station Expansion Torres, J., Jiménez, E., Soriguera, F.,

    work page doi:10.1287/opre.19.6.1363

  24. [24]

    Trondheim Municipality,

    doi:10.3390/su16166910. Trondheim Municipality,

    work page doi:10.3390/su16166910

  25. [25]

    URL:https://www.trondheim.kommune

    Detailed Regulation: Gamle Okstadbakkan & Main Cycle Path. URL:https://www.trondheim.kommune. no/aktuelt/kunngjoring-arealplan/arkiv-vedtatte-planer/eldre/2024/gamle-okstadbakkan-hovedsykkelveg- detaljregulering-r20230001/. Trondheim Municipality,

    work page 2024

  26. [26]

    URL:https://www.trondheim.kommune.no/tema/bygg- kart-og-eiendom/arealplaner/kommuneplanens-arealdeldelplaner/kpa2022-2034/

    The Municipal Plan’s Area Section 2022-2034. URL:https://www.trondheim.kommune.no/tema/bygg- kart-og-eiendom/arealplaner/kommuneplanens-arealdeldelplaner/kpa2022-2034/ . Wei, J., et al.,

    work page 2022

  27. [27]

    Journal of Transport and Land Use doi:10.5198/jtlu.2023.2262

    Correlation between the built environment and dockless bike-sharing usage. Journal of Transport and Land Use doi:10.5198/jtlu.2023.2262. Yusuf, O., Breivik, M., Rasheed, A.,

    work page doi:10.5198/jtlu.2023.2262 2023

  28. [28]

    doi:10.48550/ arXiv.2511.06875, arXiv:2511.06875

    Data-Driven Mixed-Methods Framework for Cybernetic Urban Mobility Governance. doi:10.48550/ arXiv.2511.06875, arXiv:2511.06875. M.T. Wismadi et al.:Preprint submitted to Elsevier Page 28 of 28

    work page arXiv