Content uploaded by Keyvan Hosseini
Author content
All content in this area was uploaded by Keyvan Hosseini on Nov 07, 2023
Content may be subject to copyright.
Case Studies on Transport Policy 13 (2023) 101052
Available online 22 July 2023
2213-624X/© 2023 World Conference on Transport Research Society. Published by Elsevier Ltd. This is an open access article under the CC BY license
(http://creativecommons.org/licenses/by/4.0/).
Optimising shared electric mobility hubs: Insights from performance
analysis and factors inuencing riding demand
Keyvan Hosseini , Agnieszka Stefaniec , Margaret O’Mahony , Brian Cauleld
*
Trinity Centre for Transport Research, Department of Civil, Structural and Environmental Engineering, Trinity College Dublin, Dublin, Ireland
ARTICLE INFO
Keywords:
Shared mobility
Electric mobility
Mobility hubs
E-bike
E-bike sharing system
Data envelopment analysis
ABSTRACT
In order to decarbonise the transport networks, systemic change is needed. One manifestation of this trans-
formation is shared electric mobility, seeking to curtail car usage and ownership. This current case study aims to
measure and optimise the operational performance of shared electric mobility hubs (eHUBs). From the perfor-
mance results of eHUBs, one can get helpful insights to develop appropriate future planning and management
policies for improving the transport chain. Incorporating data from September 2021 to October 2022, this
research developed a novel dynamic two-stage data envelopment analysis (DEA) framework to assess the per-
formance of the eHUB network in Inverness, Scotland. In the rst stage, the DEA model computes relative ef-
ciency scores related to the operational performance of the stations. The second stage focuses on network
analysis and examining the factors that may inuence the high or low obtained performance scores. Scrupulous
analysis shows that the population in the catchment area of the eHUBs and the weather conditions (specically,
temperature) are among the most important factors inuencing riding demand. The study also nds a weak
association between eHUBs efciency and proximity to public transport stops, suggesting that electric-assist
bikes (e-bikes, pedelecs) may not strongly complement public transport, unlike bike-sharing systems. It in-
dicates that e-bikes serve rather as a standalone mode for longer journeys. The ndings of the case study can be
used to improve sustainable mobility strategies, particularly related to e-bikes in other cities and urban areas.
1. Introduction
1.1. Background
Adopting sharing economy solutions, which emphasise possessing
less and sharing more, is one of the key strategies to tackle over-
consumption, achieving sustainability, and mitigating emissions
(Schanes et al., 2016; Miramontes et al., 2017). Shared mobility is rec-
ognised as a sharing solution to substitute for private automobility in the
transportation sector (Sopjani et al., 2020; Coenegrachts et al., 2021;
Della Mura et al., 2022). This substitution may decrease trafc congestion
(B¨
osehans et al., 2021) and, if electried, can contribute positively to
reducing air pollution (Machado et al., 2018) and GHG emissions (Martin
and Shaheen, 2016; Cauleld and Kehoe, 2021) in urban settlements.
Shared mobility can also be a benecial tool for the progressive redis-
tribution of wealth to the more vulnerable parts of society underserved by
existing transport provisions (Hosseini and Stefaniec, 2023).
Shared mobility hubs are increasingly gaining popularity and in-
vestment at research and policy levels (Rongen et al., 2022). These hubs
offer integrated multimodal shared transport services and facilitate
intermodal transfers by providing an array of mobility options in prox-
imity (Miramontes et al., 2017). What distinguishes them from free-
oating car-sharing and bike-sharing systems is the specic geograph-
ical location of a hub, making multimodal trips more convenient and
creating a sense of a designated place for travel. A xed site can also
provide a suitable location for installing charging infrastructure where
shared electric vehicles can be recharged while parked (Liao and Cor-
reia, 2022).
Whether the added value of multimodal hubs is abundant (B¨
osehans
et al., 2023) or insubstantial compared to the monomodal car-sharing
scheme (Claasen, 2020), its potential to replace private car trips
should not be underestimated (ITF, 2018). This research focuses on
shared electric mobility hubs (eHUBs), which are on-street locations that
simultaneously build on multimodal and electric mobility services. As an
* Corresponding author.
E-mail addresses: keyvan.hosseini@tcd.ie (K. Hosseini), agnieszka.stefaniec@tcd.ie (A. Stefaniec), margaret.omahony@tcd.ie (M. O’Mahony), brian.cauleld@
tcd.ie (B. Cauleld).
Contents lists available at ScienceDirect
Case Studies on Transport Policy
journal homepage: www.elsevier.com/locate/cstp
https://doi.org/10.1016/j.cstp.2023.101052
Received 17 January 2023; Received in revised form 19 July 2023; Accepted 21 July 2023
Case Studies on Transport Policy 13 (2023) 101052
2
innovative pilot low-carbon mobility project, the eHUBS project
1
aims
to deploy various electric micro-mobility options, such as e-bikes, e-
cargo bikes, and e-scooters. Currently, pilot eHUB networks are oper-
ating in ten cities in Europe. The goal of pilot projects such as the eHUBS
project is to modify mobility behaviour in society by replacing private
car trips with trips made by electric and micro-mobility means of
transport.
This work offers a unique framework for evaluating shared electric
mobility hubs and e-bike sharing systems. The presented analysis in this
research is informative in that it assesses the eHUB network by inves-
tigating multiple indicators, which recognises efciency improvement
directions within the mobility sector. The performance of eHUBs is
examined over 12 months extending the analysis to incorporate the time
factor. Furthermore, this study is the rst to measure the performance of
an e-bike sharing system using a data envelopment analysis (DEA)
framework and explore the factors inuencing the obtained efciency
scores. The dynamic slacks-based measure DEA approach (SBM) is
selected as a suitable model to provide an informative and compre-
hensive picture of the eHUB system, which can assist mobility decision-
makers in future developments and rearrangements. additionally, this
study contributes to a topical issue, given the concerted efforts made by
many governments towards shared mobility and electrifying trans-
portation. While policymakers are often sceptical about the outcomes of
the untested new forms of mobility, this research provides evidence for
the effective implementation of eHUB networks. Through these im-
provements and implications, the transport chain can advance toward
implementing challenging goals of climate neutrality and reducing pri-
vate car usage and ownership.
1.2. Previous work
The transportation sector serves as the backbone of the modern
economy. Due to its massive energy consumption, waste production, and
emission generation, it is necessary to appraise the efciency and sus-
tainability of different segments and aspects of this sector. Since DEA’s
inception, dozens of studies have utilised it to estimate efciency and
sustainability in different parts of the transportation sector (Cavaignac
and Petiot, 2017). DEA has been applied for the sustainability assess-
ment of various transport case studies within a country (Chuti-
phongdech and Vongsaroj, 2022; Gandhi et al., 2022; Stefaniec et al.,
2020) and between several countries (Güner, 2021; Stefaniec et al.,
2021). In relation to the public transport domain, Cauleld et al. (2013)
employed DEA as an appraisal tool to examine optimal public transport
investment strategies. Suguiy et al. (2020) considered the quality of
service, satisfaction of the passengers, and operation efciency index to
measure the performance of the public transport network in 50 Brazilian
cities by using DEA. Including daily shared-bicycle ridership and the
number of shared-bicycle stations as indicators in their proposed DEA
framework, Tamakloe et al. (2021) analysed transit-oriented develop-
ment in Seoul.
Tavassoli et al. (2014) proposed a model based on the SBM to analyse
the technical efciency and service effectiveness of 11 Iranian airlines.
Gong et al. (2019) utilised an SBM-based model to compare 26 major
shipping companies. Tomikawa and Goto (2022) measured the ef-
ciency of the Japanese railway system before and after privatisation by
combining both radial and nonradial DEA. Using SBM, Quintano et al.
(2020) assessed the eco-efciency of 24 ports in Europe. In line with
other studies (Rashidi and Cullinane, 2019; Liu et al., 2017; Lee et al.,
2014; Cook et al., 2013), they conrmed that outcomes generated from
SBM are more precise and reliable compared to radial DEA models when
dealing with complex case studies. To effectively deal with complex
dynamic applications (when the system runs over separated time
periods), Tone and Tsutsui (2010) introduced the dynamic SBM model,
which investigates and evaluates DMUs in separated periods.
A few research papers also concentrate on the performance mea-
surement of shared mobility in urban settlements from different points
of view. Focusing on the efciency evaluation of Malm¨
o’s public bike-
sharing stations and their determinants, Caggiani et al. (2021)
employed DEA to identify the best-performing stations in the system. In
their model, they considered the usage trends of each station as outputs
and the characteristics of stations are among their inputs. They
concluded that their outcomes could assist service planners in reallo-
cating existing resources in the bike-sharing system. The turnover rate of
bike-sharing stations is dened as both the daily number of bikes rented
from a station divided by the same station’s capacity and the daily
number of bikes returned to a station over the same station’s capacity
(Jim´
enez et al., 2016). Aiming to improve the efciency of the bike-
sharing system in Seoul, Hong et al. (2020) utilised turnover rate and
balancing rate as outputs in their two-stage DEA-based framework. Also,
the number of bicycle racks and the ratio between the bicycle paths and
vehicle lanes are selected as inputs. They mentioned their short study
period (one month) as the main limitation of their work. In an effort to
reduce the occurrence of accidents involving micro-mobility, Prencipe
et al. (2022) employed input-oriented radial DEA to assess the safety of
urban areas. Their study was conducted in the city of Bari, Italy and
involved the consideration of various inputs including population size,
number of educational institutions, hospitals, and bus stops.
Apart from DEA, a number of studies explored electric micro-
mobility from diverse angles, using different analytical tools. For
instance, Bardi et al. (2019) constructed an ordered probit model to
identify the determinants of satisfaction levels and usage patterns in e-
bike sharing systems. De Kruijf et al. (2021) employed binary regression
analysis to investigate the relationship between e-cycling and weather
conditions in Noord-Brabant, the Netherlands. They concluded that e-
bike trips decrease at higher air temperatures. Noland (2021) analyzed
the effect of weather conditions on three shared micro-mobility modes,
namely e-scooters, e-bikes, and bicycles, in Austin, Texas, using Prais-
Winsten regression analysis. The author noted that higher average
temperatures increase the duration and distance of e-bike travel, while
lower temperatures, rain, and wind have an adverse effect on e-cycling.
1.3. Structure of the paper
The central aim of the current study is to provide quantitative
empirical evidence to handle policy-related concerns on the adoption of
shared mobility. This work develops an evaluation framework to mea-
sure the relative efciency of the eHUBs and identify the factors which
determine their performance to assist local authorities in the future
planning and management of these sites.
The rest of the research paper is organised as follows. Section 2
presents the methodology. Section 3 describes the case study, dataset,
our new proposed framework, and variables. Section 4 interprets the
study’s outcomes and discusses key factors inuencing the operational
performance of eHUBs. Section 5 concludes the paper and provides
recommendations for policymakers.
2. Methodology
To measure the performance of eHUBs, the current study employs
the SBM-DEA approach for the following reasons. DEA and stochastic
frontier analysis (SFA) are two well-established quantitative approaches
for performance assessment. However, SFA, as a parametric method, is
suitable for single-output case studies. Thus, SFA is unt to appraise
complex systems such as the eHUB network when several outputs exist.
Conversely, as a nonparametric methodology, DEA can measure relative
performance in multi-input and/or multi-output cases (Wu et al., 2016).
Charnes et al. (1978) introduced radial DEA to measure the productivity
of the decision making units (DMUs). DEA is able to produce a single
1
eHUBS project website: https://www.nweurope.eu/projects/project-
search/ehubs-smart-shared-green-mobility-hubs/.
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
3
measure of performance for each DMU with multiple inputs consumed
to produce multiple outputs. It should be noted that DEA is a relative
efciency appraisal model. This means that DEA can assist decision-
makers in learning how well a DMU performs compared to its peers.
2.1. Dynamic slacks-based measure approach
Tone (2001) developed the nonradial efciency measure in DEA
called SBM. Unlike the radial model, the SBM model deals concurrently
with the input surpluses and the output shortfalls. The inefciency
components called slack variables are deducted from the unity, giving
each DMU a performance score (Tomikawa and Goto, 2022). In addi-
tion, SBM possesses greater discriminatory power than radial models to
rank DMUs (Rashidi and Cullinane, 2019). There are n DMUs that utilise
input X= (xij) ∈ Rm×n and output Y= (yij) ∈ Rs×n. The SBM method-
ology is unit-invariant and also monotone decreasing in each slack of
input s−
i(i=1,⋯,m) and output s+
r (r=1,⋯,s) that represent input
surplus and output shortfall, respectively.
Based on the technology set, the relative efciency of the focal DMUo
may be generated from the linear program. We select output-oriented
SBM to calculate the operational performance scores of eHUBs, since
it is assumed that the regulators wish to enhance the electric mobility
system output and maximize prot. The efciency score of DMUo under
the constant returns to scale assumption is obtained by:
1
ρ
o
=max
λ,s−,s+
1+1
s(∑
s
r=1
s+
r
yro)s.t.xio =∑
n
j=1
xijλj+s−
i(i=1,⋯,m),yro
=∑
n
j=1
yrjλj−s+
r(r=1,⋯,s),λj≥0(∀j),s−
i≥0(∀i),s+
r≥0(∀r)
(1)
A DMU is classied as SBM-output-efcient if the value of
ρ
o is equal
to unity. That is equivalent to having output slacks of size zero. The
model enables us to obtain projections onto the frontier for inefcient
DMUs. Based on the optimal solutions for λ*, s−*, and s+*, the input
surplus and output deciency are computed.
To demonstrate the changes in performance over time, we apply the
modied dynamic SBM to measure the performance of each station
during each month of the 12-month sample. The selected input and
output are computed in T periods indexed t=1,⋯,T. DMUs, indicators,
and parameters are identied by subscript or superscript t, which relates
them to the term. The model does not consider any additional carry-over
activities due to the absence of link indicators in the eHUB system
dataset. The dynamic system possibility set is dened as follows:
xit ≥∑
n
j=1
xijtλt
j(i=1,⋯,m;t=1,⋯,T),yrt ≤∑
n
j=1
yrjtλt
j(r=1,⋯,s;t
=1,⋯,T),λt
j≥0(j=1,⋯,n;t=1,⋯,T).(2)
The output-oriented efciency
τ
ot for period t can be computed from
the program below (Tone and Tsutsui, 2010):
1
τ
ot
=max
λ,s−,s+,su+
1+1
T[1
s(∑
s
r=1
s+
rt
yrot)],s.t.xiot =∑
n
j=1
xijtλt
j+s−
it (i=1,⋯,m;t
=1,⋯,T),yrot =∑
n
j=1
yrjtλt
j−s+
rt (r=1,⋯,s;t=1,⋯,T),λt
j
≥0(∀j,t),s−
it ≥0(∀i,t),s+
rt ≥0(∀r,t).
(3)
Because of the reciprocal of
τ
ot, the output efciency score ranges
between zero and unity. To calculate the SBM estimates, we employ
R software.
2.2. Network representation
Network analysis is a tool that detects patterns in complex systems
such as bike-sharing systems (Wu and Kim, 2020; Xin et al., 2022;
Builes-Jaramillo and Lotero, 2022) and thus suits our case study. A
representation of the network of eHUB stations was constructed as a
graph and visualised using open-source software Gephi version 0.9.7
(Gephi, 2022). The network consists of nodes that represent the eHUB
stations, and edges that link these stations. The network is directed,
meaning that the ow of e-bikes into and out of each station was taken
into account. Round trips were also included in the analysis, ensuring a
comprehensive view of network operation. The weight of the edge
represents the bidirectional ow between a pair of stations, while the
node’s weight reects the number of round trips. The distance between
stations was not considered in this analysis. To measure the connectivity
of the stations, a degree centrality metric was applied (Wu and Kim,
2020). It is the sum of the in, out, and round trips from and to a given
station.
The application of network analysis provides a comprehensive un-
derstanding of the dynamics and usage patterns inherent to the eHUB
network. It offers a clear depiction of the system’s key strengths and
weaknesses, guiding future enhancement and effective planning for the
service providers. For instance, identifying the most active stations or
routes can provide insights for potential infrastructure upgrades or more
efcient redistribution of e-bikes across the network.
3. Case study
3.1. Study area
This case study develops a framework to evaluate the operational
performance of eHUB networks. The proposed framework is based on
dynamic DEA, which can compare units that utilise and produce the
same sets of inputs and outputs. The method is applied to evaluate the
performance of the eHUB system in Inverness (known as Hi-Bike In-
verness, developed by HITRANS company) covering the 12 months of
operation from its inception in October 2021 to September 2022. During
the study period, the network offered only one type of electric vehicle: e-
bikes.
Inverness, the capital of the Scottish Highlands, is home to almost
50,000 inhabitants (The Scottish Government, 2022). By the end of
September 2022, there were three docking eHUBs in Inverness: Inver-
ness Campus, Inverness Railway station, and NatureScot Great Glen
House station (See Fig. 1). A docking eHUB has physical docks available,
and e-bikes can recharge their batteries there (Fig. 2). The Inverness
Campus station is situated nearby several higher education institutes.
Inverness Railway is located close to the central transit node of the city,
and NatureScot Great Glen House is in proximity to a large business
centre. Also, Inverness had three virtual (dockless) eHUBs available:
Eden Court Theatre station, Inverness Leisure station, and Raigmore
Hospital station (Fig. 1). A virtual eHUB is a GPS-dened zone where e-
bikes can be taken out or returned to the system, but there are no
charging facilities. The system has another virtual station (School of
Forestry station), but we do not consider it in our DEA analysis because
of data inconsistency. The Eden Court Theatre station operates beside a
cultural centre. Inverness Leisure station is on the doorstep of a com-
munity sports centre, and Raigmore Hospital station is located within a
hospital area. This diversication in the locations of current stations can
offer helpful information about the impact of the surrounding vicinity on
the performance of each station and provide insights for choosing sites
for new stations.
3.2. Data description
For operational efciency measurement of eHUBs based on previous
literature and the available data, the capacity of each station at the
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
4
eHUB system is considered as the sole input (Fig. 3). The eHUB network
in Inverness utilises a eet of 30 e-bikes circulating across all stations.
Each virtual station accommodates up to six e-bikes, while each docking
station has a capacity for 15. Outputs include the turnover station rate
based on arrival, balancing station rate, and monthly total length of trips
per station (in km). The turnover station rate (TO) expresses the station’s
popularity as a destination point. In other words, it shows how often the
capacity of an eHUB is used by riders. Jim´
enez et al. (2016) calculated it
as the total number of daily arrivals (AR
i
) to the station i divided by the
station’s capacity (C
i
).
TOi=ARi
Ci
(4)
The balancing station rate (BL) demonstrates the daily average oc-
cupancy of each eHUB. It shows whether there is a balance between the
number of e-bikes taken out from a station and put into a station within a
day. It is dened as the ratio between daily departure (DP
i
) and daily
arrival (Hong et al., 2020) of e-bikes at each station in the eHUB
network.
BLi=1− |DPi−ARi
ARi
|(5)
Finally, to compute the monthly total length of trips per station (LT
i
)
in km, we multiply the number of monthly departures from station i by
the monthly average distance of the trips started from station i. A
summary of the descriptive statistics of the relevant data is shown in
Table 1. The current research used data from six operating eHUBs in
Inverness, covering the 12 months from October 2021 to September
2022. To evaluate the performance based on the proposed framework,
we treat each eHUB in a particular month as an eHUB-month unit;
hence, we have 72 DMUs. High construct validity is concluded for our
DEA model because the number of eHUB-month units meets the rule of
thumb (Hosseini and Steniec, 2019) that the number of DMUs should
be bigger than triple the number of variables used for the DEA analysis.
As shown in Table 1, the minimum amount of monthly total length of
trips per station travelled in the system is zero, which is the case for
Inverness Leisure station in December 2021. Tone (2002) mentioned
two possible reasons for a DMU, having an output of size zero. First, the
DMU may never produce that specic output during the studied period,
in which case the indicator must remain untouched (zero). Second, the
DMU can generate that output, but incidentally, its observed value is
Fig. 1. Map of the eHUB network in Inverness.
Fig. 2. Inverness Campus eHUB with physical docks and shared e-bikes. (Photo
credit: Highlands and Islands Enterprise). Reprinted with consent.
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
5
zero. In this case, we must change the amount of the variable from zero
to a small positive number (for instance, 0.1). December 2021 is the only
month in which the average trip distance from Inverness Leisure station
was zero. Thus, we adjusted the average distance for Inverness Leisure
station in December from zero to 0.1 (see Hosseini and Stefaniec, 2019
for an opposite case).
Moreover, correlation analysis for the selected indicators was also
calculated and is presented in Table 2. The correlation coefcients are
signicantly positive at the 1% level, showing that the indicators are
signicantly related and suitable for use in the proposed dynamic SBM
model.
4. Results and discussion
4.1. Empirical results
The core target of this study is to propose a comparative framework
to measure the operational performance of electric shared mobility hubs
and to offer quantitative evidence-based responses to policy-related
questions on adoption of shared mobility. To compare the eHUBs on
the real dataset, one specic input and three specic outputs were used
in the dynamic SBM model (Equation (3). In this study, we focus on the
eHUB network in Inverness, and during the study period, that network
offered just one type of electric vehicle: e-bikes. Using the most recent
data from the eHUBs, this research develops a dynamic two-stage DEA
framework to profoundly explore this mobility service over 12 months.
The performance scores generated from our model represent how suc-
cessfully each eHUB in the network utilises available facilities.
Furthermore, these scores are relative, indicating the performance of
each station in comparison to its peers. Therefore, low average perfor-
mance scores do not represent the low performance of the whole system
but rather the low performance of some stations compared to others
within the network.
Table 3 reports the assessment outcomes generated from the pro-
posed dynamic framework. The operational performance scores, ranging
between 0 and 1, are used to rank the 72 eHUB-month units. In the
dynamic SBM model, DMUs obtaining a score of 1 are efcient. In
general, higher efciency values represent better performance among
the DMUs. In this way, our proposed model can distinguish among
DMUs to identify the better performers. In other words, the higher their
score, the more relatively efcient they are. Conversely, a lower score
means poorer relative performance. Using arithmetic means, we also
computed the average efciency score by stations and average monthly
efciency scores to obtain the ranking (Table 3).
The following conclusions can be extracted from Table 3. Out of the
72 station-month DMUs, three were relatively efcient: Inverness
Campus in October 2021 and September 2022, and Raigmore Hospital in
November 2021. This shows our proposed model’s high discriminatory
power, which can detect inefciencies in complex transportation sys-
tems such as an eHUB network. Although Inverness Railway was not
fully efcient during any month of the study period, it has the highest
average efciency performance among the eHUBs. The Inverness
Campus station obtained the second-highest average efciency score.
The rst two more relatively efcient stations are docking stations.
Raigmore Hospital station ranked as the third-best performer among all
stations and the most efcient virtual station. Two virtual stations, Eden
Court Theatre and Inverness Leisure, obtained lower efciency scores
than their peers and ranked fourth and fth, respectively. Finally,
NatureScot Great Glen House appeared to be the most inefcient station.
Moreover, looking at time periods, Inverness’ eHUB network per-
formed better in September 2022, November 2021, and August 2022
than in other studied months. The rst six months of running the project
happened during the COVID-19 pandemic, and the other half occurred
after the pandemic restrictions were lifted in Scotland. The average
performance score of eHUBs in the pandemic period was slightly lower
(0.339) than in the post-pandemic period (0.439), however, no statis-
tically signicant difference in performance was observed between the
means of the two periods based on the results of the T-test (Banker et al.,
2010). The eHUB network started operating in September 2021, which
might be a reason for slightly lower scores during the rst few months.
For an established shared mobility network, periods of pandemic could
elicit an increased demand for such services. Wang and Noland (2021)
suggest that when public apprehension and the imposition of social
distancing detrimentally affect public transportation usage (Sogbe,
2021), alternatives such as shared bikes and other shared micro-mobility
solutions can serve as suitable substitutes for private cars. They
demonstrate that these alternatives could potentially avoid the switch
Fig. 3. Illustration of the proposed dynamic framework.
Table 1
Descriptive statistics of inputs and outputs for the proposed performance mea-
surement model.
Min Max Mean Std. Dev.
Station capacity 6.000 15.000 10.500 4.532
Turnover rate 0.005 0.753 0.174 0.163
Balancing rate 0.333 1.000 0.870 0.141
Monthly total length of trips (km) 0.000 2049.520 359.309 468.487
Table 2
Spearman correlation coefcients matrix of indicators.
Station
capacity
Turnover
rate
Balancing
rate
Monthly total
length of trips
Station capacity 1
Turnover rate
a
0.350 1
Balancing rate
a
0.395
a
0.365 1
Monthly total
length of trips
(km)
a
0.739
a
0.798
a
0.395 1
a
Correlation is signicant at 1%.
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
6
from public transportation to private automobility when possible similar
pandemics hit in future.
4.2. Exploring the inuencing factors
As an intricate system, the low or high efciency of the trans-
portation chain can stem from multiple reasons. To investigate how
factors outside the boundary of the eHUB system inuence the perfor-
mance scores, we perform a squared correlation analysis between ob-
tained scores and the number of factors. Previous research pointed out
weather conditions (De Kruijf et al., 2021; Noland, 2021), population in
the catchment area (Mateo-Babiano et al., 2016; Zhang et al., 2019), and
proximity to the public transport network (McBain and Cauleld, 2018;
Oeschger et al., 2020; Caggiani et al., 2021) as factors which may have
inuence the demand for micro-mobility sharing systems.
To evaluate the impact of weather conditions on the performance of
these 72 eHUB-month DMUs in Inverness, we calculate the correlation
coefcient between average performance scores (Monthly) and several
meteorological indicators. These weather indicators presented in
Table 4 were compiled from the data collected by the nearest meteo-
rological station in Nairn (The Meteorological Ofce, 2023). These in-
dicators include monthly average temperature (◦C), maximum
temperature (◦C), minimum temperature (◦C), number of frost days,
amount of rainfall (mm), and number of sunshine hours.
Weather conditions appear to importantly inuence the riding de-
mand at the studied eHUBs. The correlation coefcients related to
temperature are strong, positive, and signicant, representing that
higher temperatures increase riding demand at eHUBs (Table 5). This
nding is consistent with the results reported by Noland (2021), who
found that higher temperatures increase the demand for micro-mobility.
However, since the maximum recorded temperature in Inverness during
the study period was 20.1 ◦C (Table 4), we were unable to explore the
inuence of excessively high temperatures on e-bike usage. Therefore,
our observation does not contradict the ndings of De Kruijf et al.
(2021), who demonstrated that high temperatures decrease e-cycling.
The maximum temperature recorded during their study period was 33.4
◦C.
Furthermore, we nd neither a strong nor signicant correlation
related to number of frost days, the amount of rainfall, and the number
of sunshine hours per month. This may be attributed to the oceanic
climate of Inverness, characterized by a considerable level of precipi-
tation that persists throughout the year. It is worth mentioning that
there were no occurrences of frost days in October 2022, nor in the
period spanning from May 2022 to September 2022. Notably, the month
of December 2022, which exhibited the weakest performance score
(Table 3), recorded the highest number of frost days at 11 (Table 4). This
observation potentially suggests that the frequency of frost days may
have contributed to the diminished operational performance score
during December 2022.
We examine the effects of indicators such as population in the
catchment area of stations (within a radius of 400 m, a walkable dis-
tance), availability of public transportation (number of bus stops within
a radius of 1 km), number of round and one-way trips started from each
station, and weather conditions. It is worth mentioning that the popu-
lation in the catchment area was estimated based on HITRANS’ open-
access geographic information. Also, the number of bus stops in prox-
imity to each station was obtained from Google Maps.
The Spearman’s correlation coefcient between efciency scores and
population in the catchment area of stations equals 0.670, indicating a
strong signicant positive association between them (Table 6). There-
fore, the low performance of NatureScot Great Glen House and Inverness
Leisure stations could be due to the low population in their surrounding
Table 3
Performance score of eHUBs from October 2021 to September 2022.
Time period eHUBs
Eden Court
Theatre
Inverness
Campus
Inverness
Leisure
Inverness
Railway
NatureScot Great Glen
House
Raigmore
Hospital
Average
(month)
Rank
(month)
October 2021 0.243 1.000 0.200 0.554 0.082 0.338 0.403 6
November 2021 0.397 0.627 0.327 0.716 0.126 1.000 0.532 2
December 2021 0.125 0.277 0.004 0.307 0.014 0.590 0.219 12
January 2022 0.096 0.214 0.048 0.389 0.050 0.614 0.235 11
February 2022 0.033 0.160 0.126 0.353 0.016 0.784 0.245 10
March 2022 0.420 0.576 0.225 0.609 0.072 0.489 0.398 7
April 2022 0.137 0.522 0.320 0.605 0.225 0.754 0.427 5
May 2022 0.151 0.386 0.102 0.624 0.120 0.584 0.328 9
June 2022 0.236 0.607 0.081 0.678 0.117 0.464 0.364 8
July 2022 0.639 0.564 0.308 0.770 0.206 0.238 0.454 4
August 2022 0.307 0.824 0.633 0.888 0.105 0.123 0.480 3
September 2022 0.647 1.000 0.648 0.889 0.154 0.133 0.579 1
Average (station) 0.286 0.563 0.252 0.615 0.107 0.509 – –
Rank (station) 4 2 5 1 6 3 – –
Table 4
Weather indicators on a monthly basis from October 2021 to September 2022 (The Meteorological Ofce, 2023).
Average Temperature
(◦C)
Maximum Temperature
(◦C)
Minimum Temperature
(◦C)
Number of frost days per
month
Rainfall
(mm)
Number of Sunshine
hours
October 2021 10.40 13.8 7 0 70.4 61.3
November 2021 7.30 10.4 4.2 2 56.7 35.7
December 2021 4.05 6.9 1.2 11 45.9 41.9
January 2022 6.10 8.7 3.5 2 25.3 56.1
Febuary 2022 4.65 7.7 1.6 4 95.2 76.1
March 2022 6.40 11.7 1.1 10 16.1 204.9
April 2022 7.65 11.6 3.7 3 75.4 137
May 2022 11.55 15.3 7.8 0 48.6 156.5
June 2022 13.95 18.6 9.3 0 29.4 209.6
July 2022 15.85 20.1 11.6 0 43.4 143
August 2022 15.40 19.5 11.3 0 31.8 172.1
September 2022 13.35 16.7 10 0 57.5 110.4
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
7
area compared to other stations. Conversely, the relatively high scores
obtained by Inverness Railway and Inverness Campus correspond to a
higher population in their catchment area. Also, the docks at the
NatureScot Great Glen House station are tucked away from the main
road and located beside the building entrance. This may cause low
visibility and, as a result, low performance for this eHUB. Also, the In-
verness Railway station stands beside the city’s major transit point,
which can be another reason for the high efciency of this station. In the
case of the Inverness Campus station, its proximity to two universities
constitutes another reason for the high efciency scores of this station,
particularly in September and October. Given the above, we conclude
that the population in the walkable area around eHUBs is a crucial factor
to be considered when situating them.
Regarding proximity to public transport, this study found a weak but
signicant correlation coefcient (0.392) between obtained efciency
scores and the number of bus stops within a 1 km radius. However, no
signicant correlation was found when considering bus stops within a
400 m radius (Table 6). Therefore, this research is unable to conrm the
ndings of previous studies that examine bike-sharing systems and
propose that these systems complement public transport (Shaheen and
Chan, 2016; McBain and Cauleld, 2018; Caggiani et al., 2021). This
inconsistency may necessitate treating e-bike sharing systems differently
from bike sharing systems in terms of their relationship with public bus
transport. Specically, the stronger association with public transport for
bikes compared to e-bikes suggests a more complementary relation with
the former rather than the latter. This observation implies that e-bikes,
with their motor assistance, have the potential to overcome barriers
associated with traditional bike riding such as gradients and physical
effort. This enables them to cover distances that may have previously
required a combination of bikes and local public transport, though
further research is needed to conrm this assertion. Nevertheless, the
high efciency of Inverness Railway could demonstrate a positive rela-
tionship between shared e-micro-mobility and railway services, as well
as potentially other long-distance public transport modes. Situating
stations near transit hubs can also provide a traveller with a means to
cover the rst and last mile of a journey and, in doing so, contribute to a
modal shift toward sustainable transport.
4.3. Network representation of the eHUB system
Correlation analysis is also performed between the efciency scores
and the number of monthly round and one-way trips (Table 6). It in-
dicates a very strong association with respect to one-way trips and a
strong relationship to round trips. Of the total number of trips, one-way
trips constituted 62.15%, while the remaining 37.85% consisted of
round trips. This observation is signicant because some service pro-
viders in other pilot cities are currently focused exclusively on e-bike
round trips. Therefore, it is recommended that they consider including
the possibility of one-way trips in their future plans to accommodate this
popular travel preference. Additionally, excelling in providing both one-
way and round-trip options would enhance the overall usability and
appeal of the eHUB network. To more adequately investigate these ob-
servations, we perform a network analysis of the eHUB system.
A network representation of the eHUB stations provides useful in-
formation about the links between stations and the riding volume. The
graphs in Fig. 4 show the cumulative network ow in the period from
October 2021 to September 2022. The number of rides between stations
and round trips was expressed as a percentage of all trips in the network
in a given period to enable comparison over time. The number of in and
out trips were calculated, and the sum of the edges for these is reported.
The graphs show that the Inverness Railway-Inverness Campus
segment dominated the network and accounted for 66.30% of all trips
(including round trips) in the study period. Between the two stations, the
campus was more popular for round trips, which made up one-fth of all
trips taken within the network. However, the connectivity analysis
found that Inverness Railway attracted the highest number of eHUB
system users: 41.74% (Table 7).
A substantial number of journeys also used the route linking Inver-
ness Railway with Raigmore Hospital. The three points – Inverness
Railway, Inverness Campus, and Raigmore Hospital – are centrally
located in Inverness. The least popular were connections between sta-
tions located far apart, such as NatureScot Great Glen House-Raigmore
Hospital or Inverness Railway-School of Forestry. The lowest connec-
tivity was observed for Inverness Leisure, followed by NatureScot Great
Glen House and Eden Court Theatre (Table 7). School of Forestry was
not used continuously throughout the study period, hence cannot be
fully analysed.
5. Conclusions and policy implications
Using a novel dynamic DEA-based model, this research examines the
performance of a shared mobility system known as shared electric
mobility hubs (eHUBs) in Inverness, Scotland. We recognise the more
relatively efcient stations in eHUB network and the determinants of
their higher operational efciency. Identifying the features of better-
performance stations will assist service providers and policy makers in
designing and expanding the studied network and other similar mobility
systems more effectively. The ndings indicate that eHUB stations
located in close proximity to transit hubs and higher education in-
stitutions are more operationally efcient. The Inverness Railway and
Inverness Campus locations were proven to be great trip attractors, with
popularity rising on warmer days. The empirical evaluation of the
eHUBs in Inverness leads to the following conclusions and mobility
policies which can be generalised to e-bike-sharing systems in other
regions.
First, the analysis shows that the population within the walkable area
Table 5
Spearman’s correlation coefcients between Average Performance scores and the weather indicators.
Average
Temperature
Maximum
Temperature
Minimum
Temperature
Number of frost days per
month
Rainfall
(mm)
Number of Sunshine
hours
Average Performance scores
(Monthly)
0.629
a
0.580
a
0.615
a
−0.486 0.189 0.119
a
Correlation is signicant at 5%.
Table 6
Spearman’s correlation coefcients between performance scores and the selected indicators.
Population within a radius of
400 m
Bus stops within a radius
of 1 km
Bus stops within a radius of
400 m
Number of Round
trips
Number of One-way
trips
Total number of
trips
Performance
scores
0.670
b
0.392
b
0.224 0.567
b
0.843
b
0.806
b
b
Correlation is signicant at 1%.
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
8
around the eHUB stations is a critical factor in achieving optimal de-
mand for the service. Stations with a larger population in their catch-
ment area appeared to have better performance. For instance, the
Inverness Railway, with the highest population in its catchment area,
was the most efcient during the study period. Conversely, the
NatureScot Great Glen House station, located in a catchment area with
the lowest population, displayed the weakest efciency. In this study,
the catchment area was regarded as comprising only residents, but it is
important to acknowledge that travellers or tourists could also increase
the demand for e-bike services. An underlying factor that may explain
the effects of both could be the volume of pedestrian trafc; this data,
however, has not been available and, if collected, could be used in future
research.
Second, while previous research (Shaheen and Chan, 2016; McBain
and Cauleld, 2018; Caggiani et al., 2021) indicates a complementary
relationship between bike-sharing systems and public transport, this
study reveals a weak association between the performance of e-bike
stations and public bus services. This highlights the potential of e-bikes
as a standalone mode of transportation for longer journeys. Equipped
with motor support, e-bikes effectively address obstacles such as dis-
tance, gradient, and physical exertion, thereby eliminating the necessity
of combining them with local public transport for complementing
extended travel. Moreover, e-bikes are well-suited for complimenting
long-distance travel, as evidenced by the popularity of the Inverness
Railway among eHUB users. Inverness Railway was most popular among
e-HUBs users indicating that long-distance travel rather than short-
distance is complimented by e-bikes. This has an implication for plan-
ning multi-modal transport by locating e-bike sharing stations at the
transit hubs rather than local public transport stops.
Third, the study reveals that virtual eHUB stations have lower per-
formance scores than docking eHUBs. Additionally, the NatureScot
Great Glen House station, despite being a docking station, receives the
lowest efciency score. This could be due to its location being tucked
away from the main road and adjacent to the building entrance, making
it difcult to nd. Hence, the lack of recognisability for both the
NatureScot Great Glen House station and dockless stations could be a
factor contributing to their lower operational performance. Although
frequent users might more readily locate virtual stations with time, this
may pose a difculty for new and sporadic riders. It is suggested that the
relationship between station recognisability and attracting e-bike users
requires further investigation, as existing literature does not provide
sufcient evidence, and this study was not specically designed to
establish a causal relationship. Nevertheless, increasing the station’s
visibility in open-access geographic maps such as Google Maps or
placing the stations near main roads can enlarge the recognisability of
stations and may improve the system’s performance.
Fourth, the impact of temperature was found to be considerable. The
results revealed that higher temperature increases the riding demand at
eHUBs. Interestingly, the amount of rainfall and number of sunshine
hours seem not to inuence the riding patterns. The oceanic climate of
Inverness, with a substantial precipitation level that persists throughout
the year, may be a reason for this. While no signicant correlation was
found between the number of frost days and the monthly performance
score, it is notable that December 2022, with the weakest performance
score, experienced the highest number of frost days. This observation
indicates that frost days may have played a role in the low operational
performance during December 2022. On the other hand, the impact of
the COVID-19 pandemic on performance scores was not found to be
signicant. This might show that eHUBs could serve as substitutes for
private cars when the restrictions reduce public transport usage during
incoming pandemics.
Fifth, this study does not account for the quality of cycling infra-
structure in Inverness, although it is a crucial factor in attracting e-bike
users. The bicycle lanes in the city are scattered and disconnected. Also,
Inverness does not have sufcient segregated cycle infrastructure, which
affects the overall perception of the city’s attractiveness for cyclists and
e-bike users. As shown in the case of Seville, demand for shared bikes
may rise if the local government invests in improving the provision of
infrastructure following the guideline rules proven effective in the
Spanish city (Marqu´
es et al., 2015). Good-practice network design fea-
tures include the segregation of bike lanes from motorised trafc,
bidirectionality of cycling trafc, continuity of paths without gaps,
uniform design and pavement, and connecting trip attractors with res-
idential areas (Cauleld et al., 2020).
Finally, we hope that this quantitative analysis serves as a
Fig. 4. Ehub stations network represented by nodes and edges.
Table 7
Connectivity of eHUB stations, whole period.
Station All trips (%)
Eden Court Theatre 4.81
Inverness Campus 36.31
Inverness Leisure 3.12
Inverness Railway 41.74
NatureScot Great Glen House 4.66
Raigmore Hospital 7.65
School of Forestry 1.71
Total 100
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
9
groundwork for future research and practical applications, facilitating a
more rigorous examination of the interconnected implications and
mechanisms of shared mobility and electric micro-mobility in trans-
portation system development and policy formulation. This case study
was conducted based on data from a small-medium-sized urban area.
Therefore, future research could analyse the electric shared mobility
systems in larger cities and metropolitan areas. Such future research
should complement our ndings by showing diverse perspectives and
may bring further practical insights into implementing electric shared
mobility systems.
CRediT authorship contribution statement
Keyvan Hosseini: Conceptualization, Methodology, Software,
Validation, Formal analysis, Investigation, Data curation, Writing –
original draft, Writing – review & editing, Visualization, Project
administration. Agnieszka Stefaniec: Conceptualization, Methodology,
Software, Validation, Formal analysis, Investigation, Writing – original
draft, Writing – review & editing, Visualization. Margaret O’Mahony:
Conceptualization, Methodology, Validation, Investigation, Resources,
Writing – original draft, Writing – review & editing, Supervision, Project
administration, Funding acquisition. Brian Cauleld: Conceptualiza-
tion, Methodology, Validation, Investigation, Resources, Writing –
original draft, Writing – review & editing, Supervision, Project admin-
istration, Funding acquisition.
Declaration of Competing Interest
The authors declare that they have no known competing nancial
interests or personal relationships that could have appeared to inuence
the work reported in this paper.
Acknowledgements
This research was supported by Interreg North-West Europe (Project
number: NWE 826). The sponsor neither had any direct involvement in
the conduct of this work nor had any inuence on the decision to publish
this research paper. The authors wish to acknowledge Ms Vikki Trelfer
for providing data.
References
Banker, R.D., Zheng, Z.E., Natarajan, R., 2010. DEA-based hypothesis tests for comparing
two groups of decision making units. Eur. J. Oper. Res. 206 (1), 231–238. https://
doi.org/10.1016/j.ejor.2010.01.027.
Bardi, A., Mantecchini, L., Grasso, D., Paganelli, F., Malandri, C., 2019. Flexible mobile
hub for E-bike sharing and cruise tourism: a case study. Sustainability 11 (19), 5462.
https://doi.org/10.3390/su11195462.
B¨
osehans, G., Bell, M., Thorpe, N., Liao, F., de Almeida, H., Correia, G., Dissanayake, D.,
2021. eHUBs—Identifying the potential early and late adopters of shared electric
mobility hubs. Int. J. Sustain. Transp. 1–20 https://doi.org/10.1080/
15568318.2021.2015493.
B¨
osehans, G., Bell, M., Thorpe, N., Dissanayake, D., 2023. Something for every one? -an
investigation of people’s intention to use different types of shared electric vehicle.
Travel Behav Soc. 30, 178–191. https://doi.org/10.1016/j.tbs.2022.09.004.
Builes-Jaramillo, A., Lotero, L., 2022. Spatial-temporal network analysis of the public
bicycle sharing system in Medellín. Colombia. J. Transp. Geogr. 105, 103460
https://doi.org/10.1016/j.jtrangeo.2022.103460.
Caggiani, L., Camporeale, R., Hamidi, Z., Zhao, C., 2021. Evaluating the efciency of
bike-sharing stations with data envelopment analysis. Sustainability 13 (2), 881.
https://doi.org/10.3390/su13020881.
Cauleld, B., Kehoe, J., 2021. Usage patterns and preference for car sharing: a case study
of Dublin. Case Stud. Transp. 9 (1), 253–259. https://doi.org/10.1016/j.
cstp.2020.12.007.
Cauleld, B., Bailey, D., Mullarkey, S., 2013. Using data envelopment analysis as a public
transport project appraisal tool. Transp Policy. 29, 74–85. https://doi.org/10.1016/
j.tranpol.2013.04.006.
Cauleld, B., Carroll, P., Ahern, A., 2020. Transitioning to low carbon and sustainable
mobility, Working Paper No. 8, Climate Change Advisory Council, Ireland. htt
ps://www.climatecouncil.ie/media/climatechangeadvisorycouncil/contentasset
s/publications/Working%20Paper%20No.%208.pdf.
Cavaignac, L., Petiot, R., 2017. A quarter century of Data Envelopment Analysis applied
to the transport sector: a bibliometric analysis. Socio-Econ. Plan. Sci. 57, 84–96.
https://doi.org/10.1016/j.seps.2016.11.003.
Charnes, A., Cooper, W.W., Rhodes, E., 1978. Measuring the efciency of decision
making units. Eur. J. Oper. Res. 2 (6), 429–444. https://doi.org/10.1016/0377-2217
(78)90138-8.
Chutiphongdech, T., Vongsaroj, R., 2022. Technical efciency and productivity change
analysis: a case study of the regional and local airports in Thailand. Case Stud.
Transp. 10 (2), 870–890. https://doi.org/10.1016/j.cstp.2022.03.001.
Claasen, Y., 2020. Potential effects of mobility hubs: intention to use shared modes and
the intention to reduce household car ownership. University of Twente accessed 15
November 2022.
Coenegrachts, E., Beckers, J., Vanelslander, T., Verhetsel, A., 2021. Business model
blueprints for the shared mobility hub network. Sustainability 13 (12), 6939.
https://doi.org/10.3390/su13126939.
Cook, W.D., Tone, K., Zhu, J., 2014. Data envelopment analysis: prior to choosing a
model. Omega 44, 1–4. https://doi.org/10.1016/j.omega.2013.09.004.
de Kruijf, J., van der Waerden, P., Feng, T., B¨
ocker, L., van Lierop, D., Ettema, D.,
Dijst, M., 2021. Integrated weather effects on e-cycling in daily commuting: a
longitudinal evaluation of weather effects on e-cycling in the Netherlands. Transp
Res Part A Policy Pract. 148, 305–315. https://doi.org/10.1016/j.tra.2021.04.003.
della Mura, M., Failla, S., Gori, N., Micucci, A., Paganelli, F., 2022. E-scooter presence in
urban areas: are consistent rules, paying attention and smooth infrastructure enough
for safety? Sustainability. 14 (21), 14303.
Gandhi, N., Kant, R., Thakkar, J., 2022. Sustainable performance assessment of rail
freight transportation using triple bottom line approach: an application to Indian
Railways. Transp Policy. 128, 254–273. https://doi.org/10.1016/j.
tranpol.2022.08.023.
Gephi., 2022. The Open Graph Viz Platform. https://gephi.org/. (accessed 22 November
2022).
Gong, X., Wu, X., Luo, M., 2019. Company performance and environmental efciency: A
case study for shipping enterprises. Transp Policy. 82, 96–106. https://doi.org/
10.1016/j.tranpol.2018.04.008.
Güner, S., 2021. Ground-level aircraft operations as a measure of sustainable airport
efciency: a weight-restricted DEA approach. Case Stud. Transp. 9 (2), 939–949.
https://doi.org/10.1016/j.cstp.2021.04.013.
Hong, J., Tamakloe, R., Tak, J., Park, D., 2020. Two-stage double bootstrap data
envelopment analysis for efciency evaluation of shared-bicycle stations in urban
cities. Transp. Res. Rec. 2674 (6), 211–224. https://doi.org/10.1177/
0361198120918568.
Hosseini, K., Stefaniec, A., 2019. Efciency assessment of Iran’s petroleum rening
industry in the presence of unprotable output: a dynamic two-stage slacks-based
measure. Energy. 189, 116112 https://doi.org/10.1016/j.energy.2019.116112.
Hosseini, K., Stefaniec, A., 2023. A wolf in sheep’s clothing: exposing the structural
violence of private electric automobility. Energy Res. Soc. Sci. 99, 103052 https://
doi.org/10.1016/j.erss.2023.103052.
International Transport Forum (ITF), 2018. Shared mobility simulations for Dublin: case-
specic policy analysis. accessed 01 December 2022 OECD/ITF. https://www.itf-oe
cd.org/shared-mobility-dublin.
Jim´
enez, P., Nogal, M., Cauleld, B., Pilla, F., 2016. Perceptually important points of
mobility patterns to characterise bike sharing systems: the Dublin case. J. Transp.
Geogr. 54, 228–239. https://doi.org/10.1016/j.jtrangeo.2016.06.010.
Lee, T., Yeo, G.T., Thai, V.V., 2014. Environmental efciency analysis of port cities:
slacks-based measure data envelopment analysis approach. Transp. Policy. 33,
82–88. https://doi.org/10.1016/j.tranpol.2014.02.009.
Liao, F., Correia, G., 2022. Electric car-sharing and micromobility: a literature review on
their usage pattern, demand, and potential impacts. Int. J. Sustain. Transp. 16 (3),
269–286. https://doi.org/10.1080/15568318.2020.1861394.
Liu, H., Zhang, Y., Zhu, Q., Chu, J., 2017. Environmental efciency of land
transportation in China: a parallel slack-based measure for regional and temporal
analysis. J. Clean. Prod. 142, 867–876. https://doi.org/10.1016/j.
jclepro.2016.09.048.
Machado, C.A.S., de Salles Hue, N.P.M., Berssaneti, F.T., Quintanilha, J.A., 2018. An
overview of shared mobility. Sustainability. 10 (12), 4342. https://doi.org/10.3390/
su10124342.
Marqu´
es, R., Hern´
andez-Herrador, V., Calvo-Salazar, M., García-Cebri´
an, J.A., 2015.
How infrastructure can promote cycling in cities: lessons from seville. Res. Transp.
Econ. 53, 31–44. https://doi.org/10.1016/j.retrec.2015.10.017.
Martin, E., Shaheen, S., 2016. Impacts of car2go on vehicle ownership, modal shift,
vehicle miles traveled. In: And Greenhouse Gas Emissions: An Analysis of Five North
AmericAn Cities. Transportation Sustainability Research Center, UC Berkeley, p. 3.
Mateo-Babiano, I., Bean, R., Corcoran, J., Pojani, D., 2016. How does our natural and
built environment affect the use of bicycle sharing? Transp. Res. Part A Policy Pract.
94, 295–307. https://doi.org/10.1016/j.tra.2016.09.015.
McBain, C., Cauleld, B., 2018. An analysis of the factors inuencing journey time
variation in the cork public bike system. Sustain. Cities Soc. 42, 641–649. https://
doi.org/10.1016/j.scs.2017.09.030.
Meteorological, T., 2023. accessed 05 May 2023 Ofce. https://www.metofce.gov.uk/
pub/data/weather/uk/climate/stationdata/nairndata.txt.
Miramontes, M., Pfertner, M., Rayaprolu, H.S., Schreiner, M., Wulfhorst, G., 2017.
Impacts of a multimodal mobility service on travel behavior and preferences: user
insights from munich’s rst mobility station. Transportation 44 (6), 1325–1342.
https://doi.org/10.1007/s11116-017-9806-y.
Noland, R.B., 2021. Scootin’in the rain: does weather affect micromobility? Transp Res
Part A Policy Pract. 149, 114–123. https://doi.org/10.1016/j.tra.2021.05.003.
K. Hosseini et al.
Case Studies on Transport Policy 13 (2023) 101052
10
Oeschger, G., Carroll, P., Cauleld, B., 2020. Micromobility and public transport
integration: the current state of knowledge. Transp. Res. D: Transp. Environ. 89,
102628 https://doi.org/10.1016/j.trd.2020.102628.
Prencipe, L.P., Colovic, A., De Bartolomeo, S., Caggiani, L., Ottomanelli, M., 2022. In: An
Efciency Indicator for Micromobility Safety Assessment. IEEE, pp. 1–6. https://doi.
org/10.1109/EEEIC/ICPSEurope54979.2022.9854627.
Quintano, C., Mazzocchi, P., Rocca, A., 2020. Examining eco-efciency in the port sector
via non-radial data envelopment analysis and the response based procedure for
detecting unit segments. J. Clean. Prod. 259, 120979 https://doi.org/10.1016/j.
jclepro.2020.120979.
Rashidi, K., Cullinane, K., 2019. Evaluating the sustainability of national logistics
performance using data envelopment analysis. Transp Policy. 74, 35–46. https://doi.
org/10.1016/j.tranpol.2018.11.014.
Rongen, T., Tillema, T., Arts, J., Alonso-Gonz´
alez, M.J., Witte, J.J., 2022. An analysis of
the mobility hub concept in the Netherlands: historical lessons for its
implementation. J. Transp. Geogr. 104, 103419 https://doi.org/10.1016/j.
jtrangeo.2022.103419.
Schanes, K., Giljum, S., Hertwich, E., 2016. Low carbon lifestyles: a framework to
structure consumption strategies and options to reduce carbon footprints. J. Clean.
Prod. 139, 1033–1043. https://doi.org/10.1016/j.jclepro.2016.08.154.
Shaheen, S., Chan, N., 2016. Mobility and the sharing economy: potential to facilitate the
rst-and last-mile public transit connections. Built Environ. 42 (4), 573–588.
https://doi.org/10.2148/benv.42.4.573.
Sogbe, E., 2021. The evolving impact of coronavirus (COVID-19) pandemic on public
transportation in Ghana. Case Stud. Transp. 9 (4), 1607–1614. https://doi.org/
10.1016/j.cstp.2021.08.010.
Sopjani, L., Stier, J.J., Hesselgren, M., Ritz´
en, S., 2020. Shared mobility services versus
private car: implications of changes in everyday life. J. Clean. Prod. 259, 120845
https://doi.org/10.1016/j.jclepro.2020.120845.
Stefaniec, A., Hosseini, K., Xie, J., Li, Y., 2020. Sustainability assessment of inland
transportation in China: a triple bottom line-based network DEA approach. Transp.
Res. D: Transp. Environ. 80, 102258 https://doi.org/10.1016/j.trd.2020.102258.
Stefaniec, A., Hosseini, K., Assani, S., Hosseini, S.M., Li, Y., 2021. Social sustainability of
regional transportation: an assessment framework with application to EU road
transport. Socio-Econ. Plan. Sci. 78, 101088 https://doi.org/10.1016/j.
seps.2021.101088.
Suguiy, T., de Carvalho, M.F.H., Ferreira, P.A.V., 2020. Efciency versus satisfaction in
public transport: practices in Brazilian cities. Case Stud. Transp. 8 (3), 938–945.
https://doi.org/10.1016/j.cstp.2020.05.003.
Tamakloe, R., Hong, J., Tak, J., 2021. Determinants of transit-oriented development
efciency focusing on an integrated subway, bus and shared-bicycle system:
application of Simar-Wilson’s two-stage approach. Cities. 108, 102988 https://doi.
org/10.1016/j.cities.2020.102988.
Tavassoli, M., Faramarzi, G.R., Saen, R.F., 2014. Efciency and effectiveness in airline
performance using a SBM-NDEA model in the presence of shared input. J. Air
Transp. Manag. 34, 146–153. https://doi.org/10.1016/j.jairtraman.2013.09.001.
Inverness, Understanding Scottish Places (USP) The Scottish Government. 2022 accessed
16 November 2022 https://www.usp.scot/Town?mainTownName=Inverness.
Tomikawa, T., Goto, M., 2022. Efciency assessment of japanese national railways before
and after privatisation and divestiture using data envelopment analysis. Transp
Policy. 118, 44–55. https://doi.org/10.1016/j.tranpol.2022.01.012.
Tone, K., 2001. A slacks-based measure of efciency in data envelopment analysis. Eur.
J. Oper. Res. 130 (3), 498–509. https://doi.org/10.1016/S0377-2217(99)00407-5.
Tone, K., 2002. A slacks-based measure of super-efciency in data envelopment analysis.
Eur. J. Oper. Res. 143, 32–41. https://doi.org/10.1016/S0377-2217(01)00324-1.
Tone, K., Tsutsui, M., 2010. Dynamic DEA: a slacks-based measure approach. Omega. 38
(3–4), 145–156. https://doi.org/10.1016/j.omega.2009.07.003.
Wang, H., Noland, R.B., 2021. Bikeshare and subway ridership changes during the
COVID-19 pandemic in New York City. Transp Policy. 106, 262–270. https://doi.
org/10.1016/j.tranpol.2021.04.004.
Wu, C., Kim, I., 2020. Analyzing the structural properties of bike-sharing networks:
evidence from the United States, Canada, and China. Transp. Res. A: Policy Pract.
140, 52–71. https://doi.org/10.1016/j.tra.2020.07.018.
Wu, J., Zhu, Q., Chu, J., Liu, H., Liang, L., 2016. Measuring energy and environmental
efciency of transportation systems in China based on a parallel DEA approach.
Transp. Res. D: Transp. Environ. 48, 460–472. https://doi.org/10.1016/j.
trd.2015.08.001.
Xin, R., Ai, T., Ding, L., Zhu, R., Meng, L., 2022. Impact of the COVID-19 pandemic on
urban human mobility-a multiscale geospatial network analysis using New York
bike-sharing data. Cities. 126, 103677 https://doi.org/10.1016/j.
cities.2022.103677.
G. Zhang H. Yang S. Li Y. Wen Y. Li F. Liu What is the best catchment area of bike share
station? A study based on Divvy system in Chicago, USA 5th International
Conference on Transportation Information and Safety (ICTIS) 2019 1226 1232 IEEE.
10.1109/ICTIS.2019.8883774.
K. Hosseini et al.
