Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Dieses Kapitel untersucht die Faktoren, die die Verweildauer in docklosen Bike-Sharing-Systemen in Dublin, Irland, beeinflussen, indem sie räumliche Regressionsmodelle verwenden, um räumliche Abhängigkeit zu erfassen. Die Studie kommt zu dem Ergebnis, dass soziodemografische Merkmale wie die Bevölkerungszahl nach Alter und ethnischer Zugehörigkeit, die Beschäftigung nach Sektoren und der Autobesitz sowie bebaute Umgebungsvariablen wie die Anzahl der Bushaltestellen, Fahrradstationen, Bars und Restaurants sowie Bildungseinrichtungen die Verweildauer erheblich beeinflussen. Das räumliche Fehlermodell erweist sich als am besten passend, und die Studie identifiziert spezifische Zonen mit potenziellen Überangebotsproblemen oder Neupositionierungsbedürfnissen. Die Ergebnisse bieten wertvolle Einblicke in die Faktoren, die die Fahrradnutzung beeinflussen und geben praktische Empfehlungen zur Verbesserung von Bike-Sharing-Systemen. Die Studie kommt zu dem Schluss, dass die Berücksichtigung räumlicher Effekte ein besseres Modell hervorbringt und dass das Fahrradangebot in Zonen mit einem höheren Anteil an Autobesitzern gesteigert werden kann. Darüber hinaus erleben Zonen mit mehr Haltestellen des öffentlichen Nahverkehrs aufgrund eines überwiegenden Anteils an Park-and-Ride-Fahrten höhere Verweildauer.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
Cycling promotes a healthier lifestyle and alleviates traffic congestion and air pollution. The present study focuses on the dockless bike sharing system in Dublin, Ireland and identifies the socio-economic and built environment factors that affect the dwell time. The origin- destination and timestamp databases obtained from Bleeper Bike are used to estimate the dwell time. The data is analysed using Uber H3 hexagonal zones, and non-spatial and spatial regression models (both spatial lag and spatial error models) are developed. The spatial error model was found to provide a better fit to dwell time. The study identified that key factors such as the presence of public transport stations and car ownership impact dwell time significantly.
1 Introduction
Cycling is an environmentally conscious form of transport that promotes a healthier lifestyle and helps alleviate traffic congestion and air pollution [1, 2]. In the past two decades, bike-sharing systems, which allow for flexible and environmentally friendly travel, have been gaining popularity in many cities worldwide. Recently, dockless bike sharing systems that allow users to park their bike anywhere is growing across the world. Bleeper bikes [3], is the first to introduce dockless bike sharing systems in Dublin, Ireland.
Bike sharing research areas include identifying factors that influence the demand, analysing the use of bike sharing schemes to solve the last mile problem, and repositioning of bikes [4, 5]. Most earlier studies focused on docked bike sharing systems to identify the factors influencing their demand [6, 7]. These studies have found that socio-economic characteristics, built environment factors, and weather affect the demand. Recently, dockless bike sharing systems have been introduced all over the world. The analysis of these systems and identification of factors affecting their demand is still in the nascent stage, and more studies are essential.
Anzeige
Earlier studies have employed various regression models. These range from Bayesian structural time series model, geographically weighted regression, autoregressive model with exogenous variables, spatial lag model, graph regularisation technique, negative binomial regression, machine learning models, and semi-parametric methods [8]. All the above studies have modelled bike sharing trip counts. None of the studies have analysed the effect of built environment factors on dwell time, particularly in Ireland.
In this study, we have developed regression models for dwell time by spatial zones. To evaluate the presence of and consider spatial dependence in dwell time, we developed spatial lag and error models and compared them with the non- spatial model. The present study contributes to the literature by modelling dwell time, in contrast to earlier studies modelling bike trip counts. The study findings reveal various factors that affect dwell time and provide initial evidence of the need for repositioning. The findings provide valuable information that could be used to understand and improve the bike sharing systems.
The rest of the paper is organised as follows. Section 2 presents the data and descriptive statistics. Sections 3 and 4 provide details on the methodology employed and the study results. Finally, Sect. 5 presents the study conclusions.
2 Data and Descriptive Statistics
The data utilized in this study has been sourced from Bleeper, which is the first dockless bike-sharing system established in Dublin, Ireland. This system operates within a defined service area known as the “purple zone”, which is 100 km2. Users can both access and return bikes only within this zone. The origin and destination (OD) database and the timestamp database are used in this study. The timestamp database contains the location data of the bike every 5 minutes. Since these dockless bikes can be parked anywhere, for the purpose of the analysis, the study area was divided into zones using Uber H3 hexagons [9] (Fig. 1). The resolution used is 9, implying each hexagon has an area of 0.10533 km2.
Anzeige
For each zone, dwell time was calculated as the time between the destination of journey A and the time of the origin of journey B, where journeys A and B are consecutive trips made by the same bike. If the time at destination A and the time at origin B occur on different days, the dwell time was readjusted to account for the five hours (12.00 am to 5.00 am) that the bikes cannot be unlocked. In the present study, dwell times exhibited considerable variability, ranging from 0.26 to 46,234 h, with a mean value of 4,132 hours and a standard deviation of 6,248 h for the H3 zones. This is because of the variation in bike supply and usage by H3 zone. The number of bike trip origins from each zone ranged from 0 to 290. Figure 1 compares bike trips and dwell time. A few H3 zones with more bike trips have shorter dwell times and vice versa.
Fig. 1.
Study area showing the (a) bike trips and (b) dwell time (hr) variation with resolution 9 Uber hexagons. (Basemap source: Google Maps™).
We obtained the data on the socio-demographic characteristics and built environment from multiple sources. The socio-demographic characteristics such as the number of households, population by age and ethnicity, employment by sector, and car ownership were obtained from the Central Statistics Office (CSO) public datasets [10]. This information is available at a resolution of 20 m2 as small area population statistics and was remapped to estimate different variables for the H3 zones. The built environment variables were counted as the number of bus stops, bike stations, bars and restaurants, and others using their geolocations obtained from OpenStreetMaps.
3 Methods
In the present study, non-spatial linear regression models (NS), spatial lag models (SLM) and spatial error models (SEM) are developed for dwell time. The NS model is estimated using the ordinary least squares estimator. Spatial models are considered to capture spatial dependence and understand the neighbour inter- actions, utilising a spatial weights matrix. The spatial weights matrix chosen is the 6-nearest neighbour matrix, similar to [11]. This matrix differs from the contiguity weights matrix only in the case of boundary zones. SLM and SEM are estimated using the maximum likelihood estimator (Fig. 2).
Fig. 2.
Descriptive statistics of the dependent and independent variables.
Spatial Lag Models (SLM) capture the spatial correlation in each cross-sectional unit’s dependent variable as a spatially weighted average of dependent variables of other cross-sectional units. The weighted average is the spatial lag (W Y), with its parameter known as the spatial lag parameter ρ. While X is a matrix of independent variables, β is a vector of parameters to be estimated. The corresponding reduced-form version of Eq. (1), given by Eq. (2), reveals a global range of spillovers, with A−1 = (I − ρW)−1 being the spatial multiplier or the Leontief inverse [12].
The effect of neighbourhood characteristics on dwell time are presented in this section. Table 1 presents the different models developed in this study. The R2 and AIC of the models are also shown in the table. The best model is the SEM model in terms of AIC. The spatial dependence is present in both the dependent variable (SLM) and the residuals (SEM). In both the spatial and non-spatial models, the socio-demographic characteristics such as population by age and ethnicity, employment by sector, and car ownership are found to affect the dwell time. Similarly, built environment variables also affect the dwell time.
A few variables were found to be insignificant in each of the models. Dwell time did not differ for the populations aged 25–30 and 35–40 in the SLM model. Further, the proportion of white Irish employed in the building and construction, commerce and trade sectors and the proportion of households with either 1 or 2 cars are found to be not significant. The number of Luas stations also did not affect the dwell time of a zone. Similar results are also observed in the case of SEM, except for the number of households with either 1 or 2 cars, which are found to influence the dwell time.
The SEM model is only interpreted here for the sake of brevity. The results show that zones with a higher proportion of households owning either 1 or 2 cars have shorter dwell times. This is perhaps because these zones have higher usage of bikes mixed with a lower supply. Further, zones with more bus stops and rail stations tend to experience longer dwell times, perhaps because most bike trips in these zones are for park-and-ride trips that result in higher dwell times. Additionally, zones with more bike stations, bars and restaurants, and financial institutions have higher dwell time. On the other hand, zones with more educational institutions have shorter dwell times. Further, zones with a higher proportion of black Irish population aged 55–60 have shorter dwell times. This is perhaps because the younger population and black Irish tend to rely more on bikes for their mobility than other population groups.
Table 1.
Non-spatial and spatial regression model results for dwell time
Variable
NS
SLM
SEM
(Intercept)
Population by age
Age 25–30
9881.00 (6.35) 1183.91 (1.74) 6249.44 (5.20)
−15460.00 (2.30)
Age 35–40
−15400.00 (1.62)
Age 55–60
−33400 (3.36) −26637.39 (2.96) −28209.33 (2.62)
Population by ethnicity
WI
−2.41 (3.80)
BI
Employment by sector
BCT
−32.88 (4.58)
30.26 (1.50)
−23.76 (4.55)
−27.90 (4.46)
CTT
8.29 (2.56)
TCT
Car ownership
No cars
18.26 (3.63)
15.95 (6.40)
2187.65 (1.88)
20.59 (7.25)
1 car
−4906.00 (2.21)
−7101.25 (3.36)
2 cars
Built environment
Bus stops
−8817.00 (4.49)
207.20 (3.25)
178.92 (2.90)
−4591.75 (2.68)
165.70 (2.63)
Bike stations
764.20 (6.75)
787.58 (7.19)
844.99 (7.69)
Bars and restaurants
240.40 (6.50)
213.53 (5.92)
247.14 (6.73)
Educational institutions
−797.90 (2.23)
−891.32 (2.58)
−996.18 (2.94)
Financial institutions
400.70 (1.71)
581.10 (2.60)
485.74 (2.21)
Luas stations
1239.00 (1.40)
Parks
1227.00 (2.51)
1049.71 (2.21)
944.02 (2.00)
Rail stations
1615.00 (2.01)
1270.81 (1.63)
1148.50 (1.51)
Spatial parameters
ρ
NA
0.23 (5.75)
NA
λ
NA
NA
0.29 (6.09)
No of observations
Model fit
704
R2
0.64
NA
NA
5 Conclusions
The present study considers the spatial dependence in the bike dwell times by hexagonal zones. Our findings show the presence of spatial interactions in both the dependent variable and the residuals, and considering spatial effects produces a better model. The spatial error model provides the best fit. Notably, zones with a higher proportion of car ownership have a high bike usage conditional on the supply. Bike supply can be increased in these zones. Zones with more public transport stops experience higher dwell times due to a predominant share of park-and-ride trips. Further, the analysis shows specific locations exhibiting longer dwell times, indicating potential oversupply issues or a need for bike repositioning.
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the chapter's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the chapter's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.
Fan, Y., Zheng, S.: Dockless bike sharing alleviates road congestion by com- plementing subway travel: evidence from Beijing. Cities 107(102), 895 (2020)
2.
Ma, X., Yuan, Y., Van Oort, N., Hoogendoorn, S.: Bike-sharing systems’ impact on modal shift: a case study in delft, the netherlands. J. Clean. Prod. 259(120), 846 (2020)
Zhang, Y., Thomas, T., Brussel, M., Van Maarseveen, M.: Exploring the impact of built environment factors on the use of public bikes at bike stations: case study in Zhongshan, China. J. Transp. Geogr. 58, 59–70 (2017)CrossRef
5.
Usama, M., Shen, Y., Zahoor, O.: A free-floating bike repositioning problem with faulty bikes. Procedia Comput. Sci. 151, 155–162 (2019)CrossRef
6.
El-Assi, W., Salah Mahmoud, M., Nurul Habib, K.: Effects of built environment and weather on bike sharing demand: a station level analysis of commercial bike sharing in Toronto. Transportation 44, 589–613 (2017)CrossRef
7.
Ravensbergen, L., Buliung, R., Mendonca, M., Garg, N.: Biking to ride: Investigating the challenges and barriers of integrating cycling with regional rail transit. Transp. Res. Rec. 2672(8), 374–383 (2018)CrossRef
8.
Venkadavarahan, M., Joji, M.S., Marisamynathan, S.: Development of spatial econometric models for estimating the bicycle sharing trip activity. Sustain. Cities Soc. 98(104), 861 (2023)
