Pith. sign in

REVIEW 4 major objections 5 minor 102 references

Modeling E-Bike Route Choice in Washington, DC: A Path Size Logit Approach

T0 review · 4 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper claims that shared e-bike riders in Washington, DC, systematically prefer routes that minimize conflict with motor vehicles and pedestrians and that keep their ride uninterrupted, with bicycle facilities mattering most on major…

desk verdict Solid first US e-bike route choice study with SVI, but the hybrid choice set—observed routes plus one shortest path—is the real soft spot; results are plausible but the central estimates are not fully verified. read the letter →

arxiv 2608.05449 v1 pith:P7PDJ7R4 submitted 2026-08-05 stat.AP stat.OT

Yiheng Qian , Zehui Yin , Darren M. Scott , Xiang Yan This is my paper · ORCID
classification stat.APstat.OT
keywords e-bikeroutechoicePathSizeLogitGPStrajectoriesStreetViewimagerybicycleinfrastructureroadwayhierarchyWashingtonDC
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper analyzes GPS trajectories of shared e-bike trips in Washington, DC, to determine which street and streetscape features riders actually choose. It claims that e-bike users prefer routes that keep them separated from both motor vehicles and pedestrians, and that keep riding uninterrupted: protected bike lanes, off-street paths, few turns, few crossings. The paper further claims that bicycle facilities matter more on major roads than on minor roads, and that their importance grows as trips get longer. Street-level visual features, measured from Street View images, add modest explanatory power, with trees the only greenery that consistently attracts riders. If correct, the findings give infrastructure planners a concrete ranking of which route attributes change the odds that riders pick a street.

What carries the argument

The Path Size Logit (PSL) model extends multinomial logit by adding a path-size term, $\ln(PS_i)$, that discounts the utility of a route by how much it overlaps other alternatives, computed as a length-weighted sum over links of one divided by the number of alternatives using that link. The paper's choice sets are hybrid: observed aggregated routes plus a single shortest path per origin-destination pair, with route frequencies expanded into 13,118 choice occasions. This construction lets the model compare realized routes against a feasible benchmark while avoiding fully algorithmic choice set generation. The machinery also includes interactions between route length and selected attributes to test context dependence, and standardized effect sizes (coefficient times one standard deviation) to rank practical importance.

What would settle it

Take the same 13,118 DC e-bike trips and re-estimate route preferences with a link-based model that requires no explicit choice set, or with a much larger algorithmically generated choice set; if the magnitude or sign of the protected-lane, crossings, or tree coefficients changes substantially, the hybrid choice set, rather than rider behavior, was driving the results.

Watch

Extended reading notes

Core claim

Using 13,118 choice occasions from 1,672 origin-destination choice sets, each containing observed routes plus the network shortest path, a Path Size Logit model estimates how route attributes shift utility. The central finding is that e-bike route choice is governed by two themes: dedicated, low-conflict riding space and route continuity. Protected bike lanes and off-street paths attract riders; footpaths, crossings, pedestrian zones, shared paths, turns, and stairs deter them. Facility effects are strongly moderated by roadway hierarchy, with bicycle facilities mattering much more on major roads, and by trip length, with longer trips showing stronger preferences for cycling infrastructure. Among street-level visual features, trees are consistently positive, while sky visibility, building coverage, and sidewalk visibility are negative; visual features improve model fit but their standardized effects fall below those of core infrastructure.

Load-bearing premise

The model assumes that each rider's true consideration set is captured by the observed routes in their origin-destination group plus a single shortest path; if riders seriously weighed other feasible routes not in this set, all estimated preferences would be distorted.

Editorial extensions

If this is right

  • Protected bike lanes on major roads are the most attractive infrastructure attribute in the model, so network investment targeting major-road corridors should take priority for e-bike users.
  • Painted lanes and sharrows on major roads become more valuable as trip length grows, implying that long e-bike trips can be drawn onto major roads only if some bicycle facility is present.
  • Routes with crossings, shared paths, footpaths, and sidewalk exposure are consistently avoided, suggesting separated or off-street facilities are needed to keep e-bikes out of pedestrian-heavy spaces.
  • Steep slopes above 10 percent deter even electric-assist riders, while moderate slopes do not, so network planning should focus on eliminating extreme gradient segments.
  • Tree coverage is the only street-level visual feature with a consistently positive effect, indicating that shade and greenery can make routes more attractive independently of physical infrastructure.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same hybrid choice set and Path Size Logit pipeline could be applied to conventional bicycle GPS data from the same system, allowing direct within-city comparison of e-bike versus bicycle preferences rather than cross-study comparisons.
  • If the conflict-avoidance and continuity mechanism generalizes, e-bike route choice models should routinely include pedestrian-conflict exposures such as crossing density and shared-path length, which are often omitted.
  • The finding that facility value grows with trip length suggests that infrastructure investment could be assessed by trip-length-weighted usage, not just trip counts, to capture where protected facilities deliver the most benefit.
  • The negative path-size coefficient appears sensitive to how low-frequency routes are filtered, so future datasets with more trips per origin-destination pair could clarify whether this is a real behavioral signature or an artifact of sparse alternatives.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This paper estimates a Path Size Logit model of shared e-bike route choice in Washington, DC, using 2023 Capital Bikeshare GPS trajectories. It constructs a hybrid choice set from clustered observed routes plus one shortest path per OD pair and combines OSM/GIS infrastructure variables with computer-vision-derived Street View features. The full model (Model 4, Table 3) yields negative effects of length, turns, crossings, footpaths, pedestrian zones, shared paths, and stairs; positive effects of protected and painted bike lanes and off-street bike paths; and a positive effect of tree cover. The authors report roadway-hierarchy differences (larger facility effects on major roads), length-dependent interactions for major-road facilities, and standardized effect sizes. They acknowledge infrastructure endogeneity and interpret the results as associative rather than causal.

Significance. If the estimates were identified, this would be a valuable first revealed-preference e-bike route choice study for a U.S. urban area, with a large GPS sample, explicit facility-type comparisons across road hierarchies, and a novel integration of Street View imagery. Strengths include the explicit PSL framework, likelihood-ratio comparisons across specifications, the sensitivity analysis in Appendix A, and the honest acknowledgment of infrastructure endogeneity. However, the load-bearing choice-set construction makes the current estimates hard to interpret as preference parameters, so the contribution is conditional on a convincing resolution of that issue.

major comments (4)
  1. [4.3, Eq. (3)] The hybrid choice set is outcome-based: an alternative enters C because it was observed as a chosen route (and retained only if its aggregated alternative was chosen at least twice), while the only non-chosen alternative is a single shortest path per OD pair. Under standard random-utility assumptions, estimating Eq. (3) on this sample is inconsistent unless the probability of inclusion into C is known and accounted for, because inclusion depends on the outcome. Appendix A varies only the minimum route frequency threshold and never the way C is formed (e.g., generated alternatives, sampled alternatives with known weights, or a correction), so this concern is untested. Because every coefficient in Table 3 is relative to C, the abstract's central claims may reflect choice-set composition rather than e-bike preferences. In addition, the shortest path is computed on the OSM walking network with one-way restrictions removed (Section 3.2), so it may be legally or physically unridable; the observed-versus-shortest contrast could encode network feasibility rather than taste.
  2. [5.2 and Table 3] The claim that roadway hierarchy moderates the effect of bicycle facilities is based on comparing the magnitudes of separate coefficients (e.g., Major w/ protected 2.71 vs. Minor w/ protected 2.01; Major w/o bike lane -1.43 vs. Minor w/o bike lane -0.26), but no statistical test of the difference is reported. The conclusion that the gap is 'wider on major roads' requires either an interaction model with road class or a Wald test on the coefficient differences. Similarly, the length-interaction results in Figure 4 are presented only as marginal-effect curves; the interaction coefficients, standard errors, and p-values are not reported, so the claim that longer trips exhibit stronger infrastructure preferences cannot be verified.
  3. [4.3 Step 4, Table 3] The 13,118 choice occasions are distributed over only 1,672 OD groups, and each OD group contributes repeated occasions with the same choice set and identical aggregated route attributes. The conditional logit estimation treats these as independent observations, so standard errors are likely understated. The authors should report cluster-robust standard errors by OD group (or use a mixed specification). This could change inference for coefficients near significance thresholds, such as Minor w/o bike lane (p < 0.1) and several SVI variables.
  4. [5.1, Appendix A Table A.1] The text states that restricting observed routes to frequencies of at least two 'had little influence on the estimated coefficients while producing the expected negative route length coefficient.' Table A.1 does not support this: moving from Freq>=1 to Freq>=2 changes Length from 0.00 to -0.48, ln(PS) from +0.09 to -0.68, Major w/ protected from 1.46 to 2.71, and Sidewalk from -1.92 to -9.37. These are substantial changes in magnitude and sign, not a minor filter effect. The authors should revise this statement or explain the mechanism; as written, the sign flips suggest the estimates are sensitive to the rule determining which observed routes constitute alternatives.
minor comments (5)
  1. [4.3 and Table 3] Step 3 reports 13,118 observed trips, but Model 4 (Freq>=2) has 9,686 events. Please clarify the relationship between these numbers and the two different frequency filters described in Step 3 and Section 5.1.
  2. [3.3/4.1] The nearest-SVI-point assignment has an average distance of 19 m and a 95th percentile of 114 m; this measurement error is not propagated to the route-level visual variables. Consider a distance-weighted interpolation or a robustness check.
  3. [3.2] Map-matching and shortest-path computation on the OSM walking network with one-way restrictions removed should be described in terms of the proportion of paths that use legally restricted links; this would help interpret the observed-versus-shortest contrast.
  4. [5.1] The reference for the R clogit routine should be the survival package (Therneau and Grambsch), not Reid and Tibshirani, whose clogitL1 implements penalized conditional logistic regression.
  5. [Data] Provide a data availability statement or a clear statement of restrictions on the Lyft data.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: all preference coefficients are estimated from GPS route data, and the self-citations are provenance/context rather than load-bearing identifications.

full rationale

The paper's derivation chain is: GPS trajectories -> map matching -> OD aggregation and route clustering -> hybrid choice set (observed routes plus one shortest path per OD) -> path size logit likelihood -> coefficient estimates -> behavioral interpretation. No headline result is defined in terms of itself: the path size term is computed from the constructed choice set via Eq. (1), ln(PS) enters utility with an estimated coefficient (Eq. 2 and Table 3), and the route attributes are measured from the network and imagery, not fitted to the conclusions. The shortest path is an independent non-chosen benchmark, so the observed-versus-shortest contrast is not a tautology. The Sec. 4.3 concern that choice-set inclusion depends on observed choices is a potential estimation bias (choice-based sampling of alternatives), not an algebraic reduction of the outputs to the inputs; the paper itself acknowledges infrastructure endogeneity in the limitations and interprets findings as associative. Self-citations are present but not load-bearing: Yin and Scott (2026a) is cited only as the origin of the map-matching preprocessing, with manual verification reported here; Yin and Scott (2026b) and Qian et al. (2026) are used for comparison/context, not to force any parameter or forbid an alternative. No uniqueness theorem is imported, and no ansatz is smuggled in via citation. The central claims about facility types, roadway hierarchy moderation, and route-length interactions are empirical estimates from a fitted likelihood and are externally falsifiable. Accordingly, the appropriate finding is no significant circularity; the score of 2 reflects minor non-load-bearing self-citations rather than any circular derivation.

Assumptions & free parameters 5 free parameters · 6 assumptions · 0 invented entities

All parameters are estimated from data; no new theoretical entities are introduced. The central behavioral claims depend on the hand-set choice set construction thresholds and on the fitted utility coefficients. The most consequential hand-set choices are the 150 m OD grid, the 80% route overlap threshold, and the minimum frequency of 2, which materially affect the sample and the path size coefficient.

free parameters (5)
  • Utility coefficients beta for all 28 route attributes = Table 3, Model 4 (e.g., Length -0.48, Turns -0.28, Major protected 2.71, Tree 3.27, ln(PS) -0.68)
    All behavioral conclusions rest on these fitted coefficients from conditional logistic regression.
  • OD clustering grid size = 150 m
    Chosen by hand to define origin-destination pairs; affects choice set composition and sample size.
  • Route overlap aggregation threshold = 80%
    Chosen by hand to merge near-identical observed routes; affects the definition of alternatives.
  • Minimum observed route frequency = 2
    Final model drops routes observed once; sensitivity analysis shows this changes the path size and length coefficients.
  • Minimum trips per OD group = 3
    Chosen by hand to ensure enough behavioral observations per choice set.
assumptions (6)
  • standard math Random utility maximization with i.i.d. extreme value errors (Multinomial Logit foundation)
    Invoked in Section 4.2 to justify the Path Size Logit probability formula.
  • domain assumption The OSM walking network with one-way restrictions removed approximates the feasible cycling network
    Invoked in Section 3.2 for map matching; if the network is incomplete, routes and attributes are misrepresented.
  • domain assumption Observed routes plus one shortest path form an adequate choice set for consistent estimation
    Central to Section 4.3; unobserved but considered alternatives would bias coefficient estimates.
  • domain assumption The nearest Street View image represents the visual environment of each road edge
    Invoked in Section 4.1; average matching distance is 19 m and 95th percentile is 114 m, so misassignment is possible.
  • domain assumption SegFormer pretrained on ImageNet and ADE20K provides valid semantic segmentation for street view images
    Invoked in Section 3.3; no task-specific fine-tuning is performed.
  • standard math Conditional logistic regression is consistent given the constructed choice sets
    Needed for all reported coefficients and standard errors in Section 5.1.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modeling E-Bike Route Choice in Washington, DC: A Path Size Logit Approach." pith.science (2026). https://pith.science/paper/P7PDJ7R4

@misc{pith2026260805449,
  author       = {Pith},
  title        = {Pith review of: Modeling E-Bike Route Choice in Washington, DC: A Path Size Logit Approach},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/P7PDJ7R4}},
  note         = {Machine review of arXiv:2608.05449}
}
read the original abstract

Understanding e-bike route choice is essential for developing effective cycling infrastructure, yet empirical evidence remains limited. This study investigates shared e-bike route choice in Washington, DC, using Global Positioning System (GPS) trajectory data from the Capital Bikeshare system. A Path Size Logit model is estimated using a hybrid choice set consisting of observed routes and corresponding shortest paths, integrating Geographic Information System (GIS)-based infrastructure variables with computer vision-derived street-level visual features extracted from Street View images (SVI). The results indicate that e-bike riders tend to choose routes that minimize conflicts with both motor vehicles and pedestrians while maintaining travel continuity. Roadway hierarchy substantially moderates the influence of bicycle facilities, with the presence of bicycle facilities having a much greater impact on route choice along major roads than along minor roads. Longer trips also exhibit stronger preferences for cycling infrastructure. Incorporating street-level visual features improves model performance, although their effects are generally smaller than those of road infrastructure, with trees being the only greenery component showing a consistently positive effect. Standardized effect sizes further identify the most behaviorally important route attributes. These findings provide practical evidence for cycling infrastructure planning in the e-bike era.

Figures

Figures reproduced from arXiv: 2608.05449 by the authors.

Figure 1
Figure 1. Illustration of bicycle facility types: (1) Shared lane (sharrows) on R Street NE, (2) Painted lane [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Study area and e-bike route data. to use, and actually do use, are not always clearly defined. Individuals often switch between close parallel facilities that traditional matching algorithms struggle to detect. We used the Python package Leuven.MapMatching (Meert and Verbeke, 2018) to map￾match the trip GPS trajectories based on a Hidden Markov Model with non-emitting states (Newson and Krumm, 2009). The map-matchin… view at source ↗
Figure 3
Figure 3. Street View image processing. To capture the visual and perceptual qualities of the riding environment, we extracted street-level visual features from SVI using computer vision. A total of 54,601 images were sampled and collected from an online third-party provider at unique geographic locations along the candidate network. Each image was processed using SegFormer (Xie et al., 2021), a transformer-based semantic seg… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Interaction effects between route length and selected route attributes. Each panel shows the [PITH_FULL_IMAGE:figures/full_fig_p017_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

102 extracted references · 38 canonical work pages

  1. [1]

    International Journal of Computer Vision , volume=

    Semantic understanding of scenes through the ade20k dataset , author=. International Journal of Computer Vision , volume=. 2019 , publisher=

    2019

  2. [2]

    Transportation Research Part C: Emerging Technologies , volume=

    A perturbed utility route choice model , author=. Transportation Research Part C: Emerging Technologies , volume=. 2022 , publisher=

    2022

  3. [3]

    Transportation Research Part B: Methodological , volume=

    A link based network route choice model with unrestricted choice set , author=. Transportation Research Part B: Methodological , volume=. 2013 , publisher=

    2013

  4. [4]

    Transportation Research Part D: Transport and Environment , volume=

    Big data, big bias? On factors shaping transit and shared micromobility integration , author=. Transportation Research Part D: Transport and Environment , volume=. 2025 , publisher=

    2025

  5. [5]

    Bike Lane Projects , year =

  6. [6]

    Sustainability , volume=

    Electric scooter sharing and bike sharing user behaviour and characteristics , author=. Sustainability , volume=. 2020 , publisher=

    2020

  7. [7]

    Transportation Research Record , volume=

    Comparison of E-Scooter and bike users’ behavior in mixed traffic , author=. Transportation Research Record , volume=. 2025 , publisher=

    2025

  8. [8]

    Environment international , volume=

    Using deep learning to examine street view green and blue spaces and their associations with geriatric depression in Beijing, China , author=. Environment international , volume=. 2019 , publisher=

    2019

Show all 102 references
  1. [9]

    Cities , volume=

    Associations between overhead-view and eye-level urban greenness and cycling behaviors , author=. Cities , volume=. 2019 , publisher=

    2019

  2. [10]

    Sustainable Cities and Society , volume=

    Relationship between eye-level greenness and cycling frequency around metro stations in Shenzhen, China: A big data approach , author=. Sustainable Cities and Society , volume=. 2020 , publisher=

    2020

  3. [11]

    Journal of transport geography , volume=

    Why do bicyclists take detours? A multilevel regression model using smartphone GPS data , author=. Journal of transport geography , volume=. 2019 , publisher=

    2019

  4. [12]

    Journal of cleaner production , volume=

    Cycling comfort on asphalt pavement: Influence of the pavement-tyre interface on vibration , author=. Journal of cleaner production , volume=. 2019 , publisher=

    2019

  5. [13]

    22nd Swiss Transport Research Conference (STRC 2022) , year=

    Choice set generation for large-scale cycling networks , author=. 22nd Swiss Transport Research Conference (STRC 2022) , year=

    2022

  6. [14]

    Personal and Ubiquitous Computing , volume=

    Path complexity and bicyclist route choice set quality assessment , author=. Personal and Ubiquitous Computing , volume=. 2021 , publisher=

    2021

  7. [15]

    European journal of transport and infrastructure research , volume=

    Efficiency of choice set generation methods for bicycle routes , author=. European journal of transport and infrastructure research , volume=. 2014 , publisher=

    2014

  8. [16]

    KSCE Journal of Civil Engineering , volume=

    Estimation of route choice behaviors of bike-sharing users as first-and last-mile trips for introduction of mobility-as-a-service (MaaS) , author=. KSCE Journal of Civil Engineering , volume=. 2022 , publisher=

    2022

  9. [17]

    Journal of transport geography , volume=

    Spatial analysis of shared e-scooter trips , author=. Journal of transport geography , volume=. 2021 , publisher=

    2021

  10. [18]

    Portland, OR: Portland Bureau of Transportation , year=

    E-scooter findings report , author=. Portland, OR: Portland Bureau of Transportation , year=

  11. [19]

    Sustainability , volume=

    A survey of electric-scooter riders’ route choice, safety perception, and helmet use , author=. Sustainability , volume=. 2023 , publisher=

    2023

  12. [20]

    Transportation research procedia , volume=

    E-scooter route choice in Germany--using stated preference data to investigate e-scooter route choice preferences , author=. Transportation research procedia , volume=. 2023 , publisher=

    2023

  13. [21]

    Scientific Reports , volume=

    City-scale GPS data reveals impact of spatial configuration and dedicated infrastructure on e-scooter route choice , author=. Scientific Reports , volume=. 2025 , publisher=

    2025

  14. [22]

    bicycles--A field study , author=

    Differences in route choice behavior when riding shared e-scooters vs. bicycles--A field study , author=. Journal of safety research , volume=. 2024 , publisher=

    2024

  15. [23]

    Transportation Research Part A: Policy and Practice , volume=

    Modeling e-scooter route choices: Infrastructural preferences and intervention scenario analysis , author=. Transportation Research Part A: Policy and Practice , volume=. 2026 , publisher=

    2026

  16. [24]

    Journal of transport geography , volume=

    E-scooter and bike-share route choice and detours: Modelling the influence of built environment and sociodemographic factors , author=. Journal of transport geography , volume=. 2023 , publisher=

    2023

  17. [25]

    Transportation Research Part A: Policy and Practice , volume=

    Where do cyclists ride? A route choice model developed with revealed preference GPS data , author=. Transportation Research Part A: Policy and Practice , volume=. 2012 , publisher=

    2012

  18. [26]

    Journal of transport geography , volume=

    Influential factors of the route choices of scooter riders: A GPS-based data study , author=. Journal of transport geography , volume=. 2023 , publisher=

    2023

  19. [27]

    Journal of transport geography , volume=

    Influence of the built environment on E-scooter sharing ridership: A tale of five cities , author=. Journal of transport geography , volume=. 2021 , publisher=

    2021

  20. [28]

    Travel Behaviour and Society , volume=

    Comparing younger and older bike share users’ route choice: the role of streetscape and midblock conflicts , author=. Travel Behaviour and Society , volume=. 2026 , publisher=

    2026

  21. [29]

    Transportation research part F: traffic psychology and behaviour , volume=

    Cyclists’ perception of streetscape and its influence on route choice: A pilot study with a mixed-methods approach , author=. Transportation research part F: traffic psychology and behaviour , volume=. 2023 , publisher=

    2023

  22. [30]

    Transportation , pages=

    Effects of road greenery and sky openness on non-commuting cycling route choice: insights from online map navigation , author=. Transportation , pages=. 2025 , publisher=

    2025

  23. [31]

    Journal of transport geography , volume=

    Road environments and bicyclist route choice: The cases of Davis and San Francisco, CA , author=. Journal of transport geography , volume=. 2020 , publisher=

    2020

  24. [32]

    Journal of Transport Geography , volume=

    Why choose active travel over driving? Investigating the impact of the streetscape and land use on active travel in short journeys , author=. Journal of Transport Geography , volume=. 2024 , publisher=

    2024

  25. [33]

    ISPRS International Journal of Geo-Information , volume=

    How is urban greenness spatially associated with dockless bike sharing usage on weekdays, weekends, and holidays? , author=. ISPRS International Journal of Geo-Information , volume=. 2021 , publisher=

    2021

  26. [34]

    Transportation research part A: policy and practice , volume=

    How do design aspects influence the attractiveness of cycling streetscapes: results of virtual reality experiments in the Netherlands , author=. Transportation research part A: policy and practice , volume=. 2022 , publisher=

    2022

  27. [35]

    Journal of safety research , volume=

    Safety impacts of bicycle infrastructure: A critical review , author=. Journal of safety research , volume=. 2017 , publisher=

    2017

  28. [36]

    Journal of transport and land use , volume=

    Modelling route choice of Dutch cyclists using smartphone data , author=. Journal of transport and land use , volume=. 2018 , publisher=

    2018

  29. [37]

    Journal of choice modelling , volume=

    A smooth bounded choice model: Formulation and application in three large-scale case studies , author=. Journal of choice modelling , volume=. 2025 , publisher=

    2025

  30. [38]

    Journal of transport geography , volume=

    Understanding bike share cyclist route choice using GPS data: Comparing dominant routes and shortest paths , author=. Journal of transport geography , volume=. 2018 , publisher=

    2018

  31. [39]

    2025 , note =

    Tag: highway=cycleway , howpublished =. 2025 , note =

    2025

  32. [40]

    Transportation research part A: policy and practice , volume=

    A joint bicycle route choice model for various cycling frequencies and trip distances based on a large crowdsourced GPS dataset , author=. Transportation research part A: policy and practice , volume=. 2023 , publisher=

    2023

  33. [41]

    Proceedings of the National Academy of Sciences , volume=

    Bikeability and the induced demand for cycling , author=. Proceedings of the National Academy of Sciences , volume=. 2023 , publisher=

    2023

  34. [42]

    Transportation research part A: policy and practice , volume=

    Route choice modeling for cyclists on urban networks , author=. Transportation research part A: policy and practice , volume=. 2023 , publisher=

    2023

  35. [43]

    International journal of behavioral nutrition and physical activity , volume=

    The potential modal shift and health benefits of implementing a public bicycle share program in Montreal, Canada , author=. International journal of behavioral nutrition and physical activity , volume=. 2013 , publisher=

    2013

  36. [44]

    Journal of Transport Geography , volume=

    Micromobility evolution and expansion: Understanding how docked and dockless bikesharing models complement and compete--A case study of San Francisco , author=. Journal of Transport Geography , volume=. 2020 , publisher=

    2020

  37. [45]

    Transport reviews , volume=

    Shared e-scooter micromobility: review of use patterns, perceptions and environmental impacts , author=. Transport reviews , volume=. 2023 , publisher=

    2023

  38. [46]

    Journal of transport geography , volume=

    Spatiotemporal comparative analysis of scooter-share and bike-share usage patterns in Washington, DC , author=. Journal of transport geography , volume=. 2019 , publisher=

    2019

  39. [47]

    Travel behaviour and society , volume=

    Evaluating a data-driven approach for choice set identification using GPS bicycle route choice data from Amsterdam , author=. Travel behaviour and society , volume=. 2018 , publisher=

    2018

  40. [48]

    Travel behaviour and society , volume=

    A big data approach to understanding pedestrian route choice preferences: Evidence from San Francisco , author=. Travel behaviour and society , volume=. 2021 , publisher=

    2021

  41. [49]

    International Journal of Geographical Information Science , volume=

    Fast map matching, an algorithm integrating hidden Markov model with precomputation , author=. International Journal of Geographical Information Science , volume=. 2018 , publisher=

    2018

  42. [50]

    Transportation Research Part A: Policy and Practice , volume=

    Determining an efficient and precise choice set for public transport based on tracking data , author=. Transportation Research Part A: Policy and Practice , volume=. 2020 , publisher=

    2020

  43. [51]

    Transportation Research Part A: Policy and Practice , volume=

    Relevance of detailed transfer attributes in large-scale multimodal route choice models for metropolitan public transport passengers , author=. Transportation Research Part A: Policy and Practice , volume=. 2021 , publisher=

    2021

  44. [52]

    Transportation research Part A: policy and practice , volume=

    Taking the path less travelled: Valuation of denied boarding in crowded public transport systems , author=. Transportation research Part A: policy and practice , volume=. 2021 , publisher=

    2021

  45. [53]

    Transportation Research Record , volume=

    Applying branch-and-bound technique to route choice set generation , author=. Transportation Research Record , volume=. 2006 , publisher=

    2006

  46. [54]

    Transportmetrica A: Transport Science , volume=

    Route choice sets for very high-resolution data , author=. Transportmetrica A: Transport Science , volume=. 2013 , publisher=

    2013

  47. [55]

    Transportation research part C: emerging technologies , volume=

    Estimation of walking patterns in a touristic area with Wi-Fi packet sensors , author=. Transportation research part C: emerging technologies , volume=. 2021 , publisher=

    2021

  48. [56]

    Transportation research part D: transport and environment , volume=

    What type of infrastructures do e-scooter riders prefer? A route choice model , author=. Transportation research part D: transport and environment , volume=. 2021 , publisher=

    2021

  49. [57]

    Transportation , volume=

    A combined destination and route choice model for a bicycle sharing system , author=. Transportation , volume=. 2016 , publisher=

    2016

  50. [58]

    Transportation research part D: transport and environment , volume=

    Effects of e-bikes on bicycle use and mode share , author=. Transportation research part D: transport and environment , volume=. 2015 , publisher=

    2015

  51. [59]

    Journal of transport & health , volume=

    The impact of e-cycling on travel behaviour: A scoping review , author=. Journal of transport & health , volume=. 2020 , publisher=

    2020

  52. [60]

    Journal of Cycling and Micromobility Research , volume=

    Comparative study of route choice models for cyclists , author=. Journal of Cycling and Micromobility Research , volume=. 2024 , publisher=

    2024

  53. [61]

    2020 , url =

    Bicycle Facility Design Guide: Version 2 , institution =. 2020 , url =

    2020

  54. [62]

    ISPRS International Journal of Geo-Information , volume=

    Unveiling e-bike potential for commuting trips from GPS traces , author=. ISPRS International Journal of Geo-Information , volume=. 2017 , publisher=

    2017

  55. [63]

    Frontiers in public health , volume=

    Exploring built environment factors on e-bike travel behavior in urban China: a case study of Jinan , author=. Frontiers in public health , volume=. 2022 , publisher=

    2022

  56. [64]

    Cycling was never so easy!

    “Cycling was never so easy!” An analysis of e-bike commuters' motives, travel behaviour and experiences using GPS-tracking and interviews , author=. Journal of transport geography , volume=. 2017 , publisher=

    2017

  57. [65]

    Journal of Transport Geography , volume=

    Route choice of bike share users: Leveraging GPS data to derive choice sets , author=. Journal of Transport Geography , volume=. 2021 , publisher=

    2021

  58. [66]

    Transportmetrica A: Transport Science , volume=

    Performance evaluation of choice set generation algorithms for analyzing truck route choice: insights from spatial aggregation for the breadth first search link elimination (BFS-LE) algorithm , author=. Transportmetrica A: Transport Science , volume=. 2020 , publisher=

    2020

  59. [67]

    1985 , publisher=

    Discrete choice analysis: theory and application to travel demand , author=. 1985 , publisher=

    1985

  60. [68]

    IET Intelligent Transport Systems , volume=

    Evaluation of map-matching algorithms for smartphone-based active travel data , author=. IET Intelligent Transport Systems , volume=. 2023 , publisher=

    2023

  61. [69]

    2018 , url =

    Portland Bureau of Transportation (PBOT) , title =. 2018 , url =

    2018

  62. [70]

    Annals of Operations Research , volume=

    Evaluation of choice set generation algorithms for route choice models , author=. Annals of Operations Research , volume=. 2006 , publisher=

    2006

  63. [71]

    Journal of choice modelling , volume=

    Route choice modeling: past, present and future research directions , author=. Journal of choice modelling , volume=. 2009 , publisher=

    2009

  64. [72]

    Transportation research part D: transport and environment , volume=

    A spatiotemporal analysis of e-scooters’ relationships with transit and station-based bikeshare , author=. Transportation research part D: transport and environment , volume=. 2021 , publisher=

    2021

  65. [73]

    Accident Analysis & Prevention , volume=

    Are bicycle lanes effective? The relationship between passing distance and road characteristics , author=. Accident Analysis & Prevention , volume=. 2021 , publisher=

    2021

  66. [74]

    Transportation Research Record , volume=

    Influence of bike lane buffer types on perceived comfort and safety of bicyclists and potential bicyclists , author=. Transportation Research Record , volume=. 2015 , publisher=

    2015

  67. [75]

    Lessons from the Green Lanes: Evaluating protected bike lanes in the US , author=

  68. [76]

    International Conference on Geographic Information Science , pages=

    Route choice decisions of E-bike users: Analysis of GPS tracking data in the Netherlands , author=. International Conference on Geographic Information Science , pages=. 2019 , organization=

    2019

  69. [77]

    Transportation Research Part D: Transport and Environment , volume=

    The E-Bike Potential: Estimating regional e-bike impacts on greenhouse gas emissions , author=. Transportation Research Part D: Transport and Environment , volume=. 2020 , publisher=

    2020

  70. [78]

    2018 , month =

    District of Columbia Department of Transportation and Capital Bikeshare , title =. 2018 , month =

    2018

  71. [79]

    Transport Reviews , volume=

    Shared e-scooter micromobility: review of use patterns, perceptions and environmental impacts , author=. Transport Reviews , volume=. 2023 , publisher=

    2023

  72. [80]

    Journal of Transport Geography , volume=

    Spatiotemporal comparative analysis of scooter-share and bike-share usage patterns in Washington, DC , author=. Journal of Transport Geography , volume=. 2019 , publisher=

    2019

  73. [81]

    Journal of Transport Geography , volume=

    Influence of the built environment on E-scooter sharing ridership: A tale of five cities , author=. Journal of Transport Geography , volume=. 2021 , publisher=

    2021

  74. [82]

    Transportation Research Part D: Transport and Environment , volume=

    What type of infrastructures do e-scooter riders prefer? A route choice model , author=. Transportation Research Part D: Transport and Environment , volume=. 2021 , publisher=

    2021

  75. [83]

    Regularization Paths for Conditional Logistic Regression: The

    Stephen Reid and Rob Tibshirani , journal =. Regularization Paths for Conditional Logistic Regression: The. 2014 , volume =

    2014

  76. [84]

    doi:10.5281/zenodo.20836439 , url =

    Qian, Yiheng and Lyu, Duanya and Zhang, Wenwen and Hankey, Steve and Qi, Meng and Yan, Xiang , title =. doi:10.5281/zenodo.20836439 , url =

    doi

  77. [85]

    Accident Analysis & Prevention , volume=

    Risky riding: Naturalistic methods comparing safety behavior from conventional bicycle riders and electric bike riders , author=. Accident Analysis & Prevention , volume=. 2015 , publisher=

    2015

  78. [86]

    Transportation Research Part A: Policy and Practice , volume=

    Cycling route choice preferences: A taste heterogeneity and exogenous segmentation analysis based on age, gender, Geller typology, and e-bike use , author=. Transportation Research Part A: Policy and Practice , volume=. 2025 , publisher=

    2025

  79. [87]

    Transportation Research Part A: Policy and Practice , volume=

    Modelling the impact of cycle superhighways and electric bicycles , author=. Transportation Research Part A: Policy and Practice , volume=. 2021 , publisher=

    2021

  80. [88]

    Topography Spot Elevations , year =

  81. [89]

    Transport Reviews , volume=

    E-bikes in the Mainstream: Reviewing a Decade of Research , author=. Transport Reviews , volume=. 2016 , publisher=

    2016

  82. [90]

    2021 , institution=

    E-Bikes’ Effect on Mode and Route Choice: A Case Study of Richmond, VA Bike Share , author=. 2021 , institution=

    2021

  83. [91]

    Journal of Cycling and Micromobility Research , volume=

    Bike users’ route choice behaviour: Expectations from electric bikes versus reality in Greater Helsinki , author=. Journal of Cycling and Micromobility Research , volume=. 2024 , publisher=

    2024

  84. [92]

    Transportation Research Record , volume=

    Review of current practice and research on E-bikes in transport models , author=. Transportation Research Record , volume=. 2023 , publisher=

    2023

  85. [93]

    Journal of Transport Geography , volume=

    Analysing the influence of electric bike types on route preferences in urban cycling environments: A GPS data-driven approach , author=. Journal of Transport Geography , volume=. 2025 , publisher=

    2025

  86. [94]

    Alvarez and Ping Luo , title =

    Enze Xie and Wenhai Wang and Zhiding Yu and Anima Anandkumar and Jose M. Alvarez and Ping Luo , title =. CoRR , volume =. 2021 , url =. 2105.15203 , timestamp =

    2021 arXiv

  87. [95]

    2018 E-Scooter Findings Report , year =

    2018

  88. [96]

    Transportation Research Part D: Transport and Environment , volume=

    A spatiotemporal analysis of e-scooters’ relationships with transit and station-based bikeshare , author=. Transportation Research Part D: Transport and Environment , volume=. 2021 , publisher=

    2021

  89. [97]

    European Conference on Data Analysis (ECDA), Date: 2018/07/04-2018/07/06, Location: Paderborn, Germany , year=

    HMM with non-emitting states for Map Matching , author=. European Conference on Data Analysis (ECDA), Date: 2018/07/04-2018/07/06, Location: Paderborn, Germany , year=

    2018

  90. [98]

    Proceedings of the 17th ACM SIGSPATIAL international conference on advances in geographic information systems , pages=

    Hidden Markov map matching through noise and sparseness , author=. Proceedings of the 17th ACM SIGSPATIAL international conference on advances in geographic information systems , pages=

  91. [99]

    Travel Behaviour and Society , volume=

    Decoding the first-and-last mile: Analyzing the transit connecting bike share GPS routes in Hamilton, Ontario with spatiotemporal distance decay , author=. Travel Behaviour and Society , volume=. 2026 , publisher=

    2026

  92. [100]

    Cities , volume=

    Pedal preferences: GPS-based panel data insights into bike share traffic flow across membership groups , author=. Cities , volume=. 2026 , publisher=

    2026

  93. [101]

    Journal of Transport Geography , volume=

    Understanding the route choice preferences of private and dock-based public bike users using GPS data in Seoul, South Korea , author=. Journal of Transport Geography , volume=. 2024 , publisher=

    2024

  94. [102]

    International Journal of Applied Earth Observation and Geoinformation , volume=

    A personalized bikeability-based cycling route recommendation method with machine learning , author=. International Journal of Applied Earth Observation and Geoinformation , volume=. 2023 , publisher=

    2023

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.