-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
1/11
A multiattribute customer satisfaction evaluation approach for
railtransit network: A real case study for Istanbul, Turkey
Erkan Celik, Nezir Aydin n, Alev Taskin Gumus
Department of Industrial Engineering, Yildiz Technical
University, Besiktas, Istanbul 34349, Turkey
a r t i c l e i n f o
Article history:
Received 30 December 2013
Received in revised form1 July 2014
Accepted 25 September 2014
Keywords:
Customer satisfaction
Rail transit network
SERVQUAL
VIKOR
Interval type-2 fuzzy sets
Istanbul
a b s t r a c t
Rail transit is one of the most important public transportation
types, especially in big and crowded cities.
Therefore, getting a high customer satisfaction level is an
essential task for municipalities and govern
ments. For this purpose, a survey is conducted to question the
attributes related to rail transit network(metros, trams, light
rail and funicular) in Istanbul. In this study, we present a novel
framework which
integrates statistical analysis, SERVQUAL, interval type-2 fuzzy
sets and VIKOR to evaluate custome
satisfaction level for the rail transit network of Istanbul.
Level of crowdedness and density in the trainair-conditioning
system of trains' interior, noise level and vibration during the
journey, and phone ser-vices are determined as the attributes need
improvements. On the other hand, different improvement
strategies are suggested for the rail transit network. The
proposed approach provides directions for thefuture investments and
can be generalized and applied to complex decision making problems
encounter
inexact, indenite and subjective data or uncertain
information.& Elsevier Ltd. All rights reserved
1. Introduction and related literature
Assuring a high customer satisfaction (CS) level in
publictransportation (PT) systems is an important task for the
managers
and authorities. PT providers need to evaluate the performance
of
their service quality (SQ) to determine how effective and
adequate
their service is (Hassan et al., 2013). Also,Hassan et al.
(2013)note
that the performance evaluation is needed to consider existing
and
forecasted demand trends, top activities, concerns of
stakeholders
and unmet service needs. Performance evaluation attributes
can
be used for measuring economic performance, connecting the
service provider's output and encounters, and improving the SQ
of
the organization (Transportation Research Board, 2003).SQ can be
evaluated by considering customer perceptions and
expectations. Filipovi et al. (2009)present a comparative
analysis
of the expected and perceived SQ of Belgrade public
transportation
within the period 2005 and 2007. Also,De Oa et al.
(2013)presents
a structural equation approach to evaluate the quality of
service
perceived by users of a bus transit service. SQ consists both
objec-
tive and subjective attributes as inEboli and Mazzulla (2011).
CS is
determined based on the perception of the customers
(Tyrinopoulos and Antoniou, 2008;Eboli and Mazzulla,
2009,2011)
on the SQ considering multi-attributes. Eboli and Mazzulla
(2008)
propose a multinomial logit model to measure SQ in PT. The
proposed model identies the importance of SQ attributes on
globaCS and calculates an SQ index. Mouwen and Rietveld
(2013)con-
sider multi-attributes to examine whether competitive
tenderingimproves CS for public transport or not in Netherland.
Service fre-quency, on-time performance, travel speed, and vehicle
tidiness aredetermined as the most effective attributes on
satisfaction in thetendered regions. Waiting time, cleanliness and
comfort are speci-
ed as the most valuable PT variables in (DellOlio et al.,
2011)Hassan et al. (2013) juxtapose the most common attributes
otransit service as the reliability, frequency, capacity, price,
cleanli-ness, comfort, security, staff, information, and the
ticketing system
They also add loading/ridership, travel time, travel distance
andservice duration as efciency indicators. Also, Redman et
al(2013)investigate seven improvement attributes as reliability,
fre-quency, price, speed, access, comfort, and convenience.Gerek et
al
(2004)evaluate three alternative rail transit network (RTN)
based
on four main attributes that are dened as nancial,
economicsystem planning, and policy.
Multi-Attribute Decision Making (MADM) is preferred for per-
formance analyses and evaluation of the services when
multi-at-tributes are considered to determine CS level. Many MADM
ap-proaches are applied to evaluate service quality of PT and
it
performance. Such as,Nathanail (2008)develops a framework
tomeasure the quality of services provided to the passengers
foHellenic railways. MADM procedures are exible to be combinedboth
with other MADM and mathematical modeling approaches
As an example, Zak (2011) presents the ELECTRE III method
andmulti-objective mathematical programming to evaluate mass
Contents lists available atScienceDirect
journal homepage: www.elsevier.com/locate/tranpol
Transport Policy
http://dx.doi.org/10.1016/j.tranpol.2014.09.005
0967-070X/& Elsevier Ltd. All rights reserved.
n Corresponding author. Phone: 90 212 383 3029.
E-mail address: [email protected](N. Aydin).
Transport Policy 36 (2014) 283293
http://www.sciencedirect.com/science/journal/0967070Xhttp://www.elsevier.com/locate/tranpolhttp://dx.doi.org/10.1016/j.tranpol.2014.09.005mailto:[email protected]://dx.doi.org/10.1016/j.tranpol.2014.09.005http://dx.doi.org/10.1016/j.tranpol.2014.09.005http://dx.doi.org/10.1016/j.tranpol.2014.09.005http://dx.doi.org/10.1016/j.tranpol.2014.09.005mailto:[email protected]://crossmark.crossref.org/dialog/?doi=10.1016/j.tranpol.2014.09.005&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.tranpol.2014.09.005&domain=pdfhttp://crossmark.crossref.org/dialog/?doi=10.1016/j.tranpol.2014.09.005&domain=pdfhttp://dx.doi.org/10.1016/j.tranpol.2014.09.005http://dx.doi.org/10.1016/j.tranpol.2014.09.005http://dx.doi.org/10.1016/j.tranpol.2014.09.005http://www.elsevier.com/locate/tranpolhttp://www.sciencedirect.com/science/journal/0967070X
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
2/11
transit system and optimize the crew size in
Czestochowa/Poland.Furthermore, Awasthi et al. (2011) integrate the
SERVQUAL andTOPSIS to evaluate the SQ of Montreal metro
services.Hassan et al.(2013)analyze 12 operating routes in Abu
Dhabi city for the pro-
posed multi-level framework based on a multi-attribute
evalua-tion procedure that involves weighted scoring techniques,
TOPSISandk-means clustering methods. The proposed framework is
usedto evaluate the public transit service performance at system
and
route levels. Celik et al. (2013) propose an integrated GRA
andTOPSIS approach based on type-2 fuzzy sets to improve CS in
PTservices. They compare four different public transport rms;
thebus rapid transit (BRT), IETT, private PT and Otobus Inc., in
Istanbul.
In this study, we analyze the PT in Istanbul to determine
theexisting CS level of the RTN and provide suggestions to
increasethis level as studied in Beiro and Sarseld Cabral (2007).
Theypresent a qualitative study involving PT users and car users.
Hence,
a detailed interview is conducted in the region of Porto. To
in-crease PT usage, the service should accommodate the levels
ofservice required by customers and attracts potential users. In
Is-tanbul, more than40% of the population of the Istanbulers
prefers
PT. A large amount of people has private cars, but they may
preferPT because of trafc congestion on roads. Therefore, the PT
usage
rate is high. Also, the usage rate of PT in Istanbul is higher
than thenational average. Even usage rate of PT is high; people are
still
caught by the trafc congestion, especially during the rush
hours.In our study, we distinguish crowdedness and trafc congestion
bythe time and place they occur. Crowdedness refers to the
pas-sengers while trafc congestion refers the jam on roads for
cars.
With this point of view, we aimed to state that, conversely to
othermodes of public transportations (bus, jitney, shuttle, etc.)
no trafccongestion occurs on way when passengers use the RTN. On
theother hand, crowdedness occurs in both rail and other modes
of
public transportation during the rush hours, in Istanbul. In
brief,both trafc congestion and crowdedness occur on public
trans-portation modes while only crowdedness occurs in RTN.
Thus,
passengers prefer rail transit. Since RTN (metros, trams, light
railand funicular) is a vital urban transportation in Istanbul, CS
be-comes a very important factor for managers and decision
makers.
The objective of this study proposes an integrated approach,
which includes SERVQUAL, statistical analysis, type-2 fuzzy
setsand Vlsekriterijumska Optimizacija I Kompromisno Resenje
(VI-KOR) methods, to evaluate the CS for the RTN of Istanbul. By
usingthese four methods together and in an integrated way, the
data
implying CS levels and attributes evaluations can be collected
andquantied in a healthy manner. The attributes are determinedfrom
the survey and data using statistical analysis and SERVQUALmethods.
The statistical analysis provides researchers to synthe-
size raw data and present valued information for additional
ana-lysis. SERVQUAL is proposed byParasuraman et al. (1988)as one
ofthe best evaluation methods for assessing the expectations
andperceptions (Chou et al., 2011). Then, this data is transformed
to
linguistic variables by using interval type-2 fuzzy sets
principles tomake the evaluation process more realistic. Interval
type-2 fuzzysets are more suitable, exible and intelligent than
type-1 fuzzysets to represent uncertainties for handling fuzzy
group decision
making problems (Mendel et al., 2006;Lee and Chen, 2008;Chenand
Lee, 2010; Celik et al., 2013). We then combine VIKOR withinterval
type-2 fuzzy sets to gain the rankings of the RTN. Themajor
advantages of the VIKOR method are that it can trade off themaximum
group utility of the majority and the minimum of
the individual regret of the opponent, and the calculations
aresimple and straightforward. Combining VIKOR method and inter-val
type-2 fuzzy sets is an interesting and important research to-pic.
In brief, integrating all these four methods provides a valid
and
reliable evaluation of CS level.
The rest of the paper is structured as follows. Section 2
de-
scribes the proposed methodology. In Section 3 collected
data
used on the study is introduced and the application of the
CS
evaluation is presented. The discussion and conclusion of the
pa-
per are considered in the last section.
2. The proposed methodology
In this section, rstly the dimensions of SERVQUAL are dened
(Parasuraman et al., 1988;Awasthi et al., 2011). Then the
proposed
integrated SERVQUAL and VIKOR approach based on interval
type-
2 fuzzy numbers is presented.
2.1. SERVQUAL
SERVQUAL is a valuable tool for executing analysis where a
gap
is measured as the difference between the customer
expectations
and customer perceptions. The metrics of SERVQUAL are
concisely
compiled as follows (Parasuraman et al., 1988; Awasthi et
al.,
2011):
Tangibles comprise the physical appearance of the service
fa-cility, tools, staff, and communication resources. For
instance,
appearance of metro stations, public phones, etc. Reliability
represents the ability of the service provider to
execute the promised service reliably and accurately. For
in-
stance, on time departures and arrivals of metros (trams).
Responsiveness shows the willingness of the service provider
(s) to be helpful and provide service immediately. For
instance,availability of service staff when needed.
Assurance relates to the knowledge and politeness of the
staffand their capability to reveal trust, faith and condence.
Forinstance, communication of employees' during an
emergencysituation.
Empathy denotes care and personalized attention of the
em-ployees to customers. For instance, assisting elders or
peoplewith children passing toll gates to get on station.
2.2. VIKOR based interval type-2 fuzzy sets
The Vlsekriterijumska Optimizacija I Kompromisno Resenje
(VIKOR) method is proposed as a MADM technique based on
compromise solution (Opricovic, 1998,Opricovic and Tzeng,
2004,
Tzeng et al., 2005). It provides a maximum group utility for
the
majority and a minimum of an individual regret for the
opponent.
Interval-valued fuzzy sets (Vahdani et al., 2010),
interval-valuedfuzzy with gray relational analysis (Kuo, 2011),
triangular in-
tuitionistic fuzzy numbers (Wan et al., 2013), and 2-tuple
fuzzy
numbers (Ju and Wang, 2012) are integrated with VIKOR.
Hence,
type-2 fuzzy sets reect more uncertainty than type-1 fuzzy
sets
with additional degrees of freedom (Chen and Lee, 2010). In
this
paper, the extended VIKOR method with interval type-2 fuzzy
sets
is proposed to obtain the best CS level of RTN based on
average
and the worst group scores among the set of alternatives. In the
CS
evaluation process of RTN, it is assumed that there are m
alter-
natives (rail lines), where R R R, , ... , }m1 2 , n
attributes
A A A, , ... , }n1 2 and L customers C C C C, , ... , }L1 2 .The
steps of the VIKOR, based on interval type-2 fuzzy sets, are
presented as follows:
E. Celik et al. / Transport Policy 36 (2014) 283293284
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
3/11
Step 1. The importance weights of the attributes are
calculatedusing Eq.(1).
= =
A a
A
A
A
a
a
a
...
(1)
n jk
nx
n
k
k
nk
1
1
2
1
2
where =
a ( )ja a a
L
j j jL1 2
, aj is an interval type-2 fuzzy set
i m j n k L1 , 1 , 1 and L denotes the number ofcustomers.
Step 2. The average fuzzy performance values of RTN are
alsocalculated using Eq.(2).
= =
R R R
E e
A
A
A
e e e
e e e
e e e
( )
(2)
m
c ijk
n m
n
k km
k
k km
k
nk
nk
nmk
1 2
1
2
11 12 1
21 22 2
1 2
where = e e e e L( / )ij ij ij ijL1 2
, eij is an interval type-2 fuzzy set
i m j n k L1 , 1 , 1 and L denotes the number ofcustomers.
Step 3. The weighted type-2 fuzzy decision matrix is
calculated
as follows:
= V v (3)ij mxn
where
=
=
v a e
f f f f H F H F f f f f H F H F, , , ; , , , , , ; ,
ij j ij
iU
iU
iU
iU
iU
iU
iL
iL
iL
iL
iL
iL
1 2 3 4 1 2 1 2 3 4 1 2
Step 4. The positive ideal solution * *P P( , )e v and negative
ideal
solution ( Ne ) for upper and lower reference points of the
in-terval type-2 fuzzy numbers are calculated (Kuo and
Liang,2012).
*
*
*
*
= =
=
= =
=
= =
=
* * *
* * * * * * * *
* * *
* * * * * * * *
{ }
{ }(
{ }
(
(
P e e e e j Benefit
P e e e e H E H E e e e e
H E H E
P v v v v j Benefit
P f f f f H F H F f f f f
H F H F
N e e e e j Benefit
N e e e e H E H E e e e e
H E H E
, , , max
, , , ; max , max , , , , ;
max , max
, , ... , max
, , , ; max , max , , , , ;
max , max
, , ... , min
, , , ; min , min , , , , ;
min , min
eij ij ij
iij
eiU
iU
iU
iU i
UiU
iL
iL
iL
iL
iL
iL
v ij ij iji
ij
viU
iU
iU
iU
iU
iU
iL
iL
iL
iL
iL
iL
e ij ij iji
ij
eiU
iU
iU
iU i
UiU
iL
iL
iL
iL
iL
iL
1 2 3 4 1 2 1 2 3 4
1 2
1 2 3 4 1 2 1 2 3 4
1 2
1 2 3 4 1 2 1 2 3 4
1 2
Next, the average (Si) and the worst (Ri) group scores of the
CSfor each RTN is calculated.
= + = =
( )S S S i m1
2 , 1, ,
(4i
j
n
ijU
ijL
1
= + = ( )R S S i mmax
1
2 , 1, ,
(5i
jijU
ijL
where,
= + + +
+ + +
=
= * * * *
= * * * *
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )S
f f f f f f f f
e e e e e e e e
i m
,
1, ,
ijU
j
k iU
iU
iU
iU
iU
iU
iU
iU
k iU
iU
iU
iU
iU
iU
iU
iU
1
4 14
1 4
2
2 3
2
3 2
2
4 1
2
1
4 14
1 42
2 32
3 22
4 12
= + + +
+ + +
=
= * * * *
= * * * *
( ) ( ) ( ) ( )
( ) ( ) ( ) ( )S
f f f f f f f f
e e e e e e e e
i m
,
1, ,
ijL
j
k iL
iL
iL
iL
iL
iL
iL
iU
k iL
iL
iL
iL
iL
iL
iL
iL
1
4 14
1 4
2
2 3
2
3 2
2
4 1
2
1
4 14
1 42
2 32
3 22
4 12
Step 5. The Qi is calculated according to the Siand R i using
Eq(6).
= *
*+
*
* *
( )( )
( )( )
Q vS S
S Sv
R R
R R(1 )
(6
i
i i
where * = = *= = S S S S R R R Rmin , max , min , maxi
ii
ii
ii
i, v [0, 1] is
the weight of the decision making strategy of the majority o
attributes (or maximum group utility). Then the smallest Qi
isdetermined as a compromise solution if two conditions
areacceptable.
Condition 1. The acceptable advantage: Q Q DQ R R2 1 , where= DQ
m1/( 1)
Condition 2. Acceptable stability in decision-making
QR1alternative must also be the best ranked S Rand/or .
If Condition 1 and Condition 2 are not acceptable, then the
compromise solution stays same.
3. Customer satisfaction for rail transit network: a real
case
study for Istanbul
In this section, we rst introduce the RTN infrastructure in
Is-tanbul. Later, the related information about the CS survey is
de-
scribed, and the proposed approach is applied by considering
the
attributes provided by the CS survey. Lastly, the sensitivity
ana-
lyses are presented.
3.1. Rail transit network infrastructure of Istanbul
As the most crowded city of Turkey, Istanbul has been the
most
important residential area culturally, economically,
historically and
strategically. Istanbul acts as a bridge between Europe and
Asia
continents. Based on the data obtained from Turkish
StatisticaInstitute (TUIK, 2013), the population of Istanbul was
13,854,740 in
2012, and Istanbul has the highest population density as
well
Above all according to TUIK, Istanbul's population has a high
in-
cremental rate and the population is projected to be more
than15 M and16 M by the years of 2019 and 2023,
respectivelyConsidering the rise in living standards, the people
living in Is-
tanbul expect a better public transportation for buses, BRTs
and
metros.
E. Celik et al. / Transport Policy 36 (2014) 283293 285
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
4/11
The long distance passengers prefer private cars because
tra-
veling by PT needs multi-interchanges, in Istanbul. In addition,
el-
ders, handicapped, and passengers who transport with baby
prefer
private cars because PT vehicles (bus, jitney, car etc.) are
mostly not
well designed for them. For instance, lifting to get on buses or
jit-
neys, room for wheelchair(s) and stroller (s) on jitneys are the
main
problems tend passengers to use either private cars or RTN.So
far metros and trams are seen as the most convenient way to
transport for Istanbulers. However, the length of lines in total
wasabout100 km, in 2012. Since Istanbulers still are suffering from
the
trafc congestion, it is clear that this length is not enough.
Thus,
Istanbul Metropolitan Municipality (IMM) has projects for
in-
creasing the number of lines and/or lengths. Besides,
Istanbul
Ulasim A.S. (IUAS-Istanbul Public Transportation Co.)
conducts
surveys to evaluate and increase CS.
IUAS is an afliate company, founded in 1988, of IMM which
op-
erates metros, trams, light rails and funiculars in Istanbul.
IUAS oper-
ates 7urban rail transit lines with, in total, about 100 km.
Winning the
UITP's (International Association of Public Transport) best
practice
award with its Zeytinburnu-Kabatas tram line in meeting high
custo-
mer demand, IUAS serves more than 1,100,000 passengers every
day.
3.2. Rail transit lines
In this study, we analyze the CS surveys, which are
performed
for the year of 2012. The survey is conducted in only ve of
the
seven rail transit lines, because two of these rail transit
lines are
not used for public transportation but for touristic and
outing
activities. These ve rail transit lines are shown in Fig. 1 as:
F1
(Taksim-Kabatas Fenicular Line), M1 (Aksaray-Ataturk Airport
LRT
Line), M2 (Sishane-Haciosman Metro Line), T1
(Bagcilar-Kabatas
Tram Line) and T4 (Topkapi-Habibler Tram Line). The detailed
information for these ve rail transit lines are presented
inTable 1.F1, the Taksim-Kabatas funicular line acts as a bridge
between
the Sishane-Haciosman metro line, Taksim-Tunnel Heritage
Tram,
IETT buses, privately-owned public buses, dolmus (jitney)
stations,Kabatas-Bagcilar tram line; IDO, cruise, ferry and seabus
piers in
Kabatas. M1 was opened in 1989 and modied in 1994, 1995,
1999,
2002 and later in 2012. M1 serves with 6 tunnels,9 over
ground,3
over ground viaduct and 4 under ground stations. M2 started
operating between Taksim and 4.Levent in 2000. Later in 2009
Sishane and Sanayi, and in 2011 Haciosman stations are added
to
the line. It also has the branch line to Seyrantepe from
Sanayi
Mahallesi station located on the main line. T1 line's rst
phase
between Sirkeci and Aksaray started operating in 1992, and
was
later extended to Topkapi and Zeytinburnu, and later to
Eminonu.
Finally, in 2006, continuous rail transportation became
available
from 4. Levent to Ataturk Airport thanks to T1 line's Kabatas
ex-
tension and Taksim-Kabatas funicular line. In 2011,
continuous
transportation from Kabatas to Bagcilar was materialized with
the
merging of the T1 line with the T2 line. Starting its operations
in
2007 and serving between Sehitlik (where it is integrated
with
Metrobus-BRT) and Mescid-i Selam, the T4 tram is operational
on
a line of15.3 km track.
Fig. 1. Istanbul rail transit network (RTN) map
(http://www.istanbul-ulasim.com.tr/en ).
E. Celik et al. / Transport Policy 36 (2014) 283293286
http://www.istanbul-ulasim.com.tr/enhttp://www.istanbul-ulasim.com.tr/en
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
5/11
3.3. Survey and data collection
Surveys are conducted by the team of IUAS in all stations of
the
ve rail transit lines between May 15, 2012 and June 3, 2012.
For
instance, 1076 of the surveys those are conducted in line M1
are
carried out in all18 stations with different frequencies.
These
frequencies are determined by using the real percentages of
pas-sengers who use those stations. In total, 4966 surveys are
carried
out: 1076in M1,1069in M2,1575in T1,1047in T4 and199 in F1.
Surveys are conducted on different days in a week and time of
theday, i.e., Monday peak hours in the morning, Saturday
afternoon.
The participants are selected as follows: once the pollster
arrives
at the stations she/he counts the number of passengers who
pass
the turnstiles and tries to carry out the survey with the 6th
pas-
senger, if the passenger is not willing to then the pollster
tries the
next passenger and so forth. Also statistical data according to
the
order in which passengers are conducted is available upon
request.
All surveys are carried out face to face with the passengers.
The40% of the conducted surveys are controlled over the phone
by
managers of the IUAS.The surveys include four main sections: (1)
Stations and tick-
eting: Station, time and ticket type. Four questions are
directed to
the passengers. (2) Train usage: where the passenger comes
from
and goes to, usage of other types of transportation, trip time
andthe frequency, days and the reason of usage. There are 8
questions
in this section. (3) Satisfaction: general satisfaction level
fromtrains, stations and the service provided, satisfaction from
the at-
tributes mentioned inTable 2. In total 42 questions are
conducted
in this section. (4) Demographic: age, sex, education, job,
income,
marriage status, disability, communication info and whether
she/
he lets the information obtained are used by the IUAS. This
section
includes 9 questions. The demographic and statistical summary
of
the data is presented inTable 2.
3.4. Weights of the attributes
In this study, SERVQUAL is applied to classify the
attributes.
Twenty-six attributes are classied under ve dimensions of
theSERVQUAL, which are assurance, empathy, reliability,
responsive-
ness and tangibles. CS attributes and related dimension
classi-
cations are presented inTable 3.The weight of each attribute is
determined based on ve ques-
tions (what is the rst, the second, the third, the fourth and
the
fth important attribute for you, to improve CS level?), which
are
asked to 4966 customers. The averages of the importance level
of
the weights, which are in linguistic terms, are converted into
type-2
fuzzy numbers using ve different scales in Table 4 via Eq. (1)
in
Step 1. Consequently, based on type-2 fuzzy numbers, the
weights
of 26 attributes are ranked. According to evaluated averages,
the
weights of top ve attributes are determined as: waiting time
for
metro before departure (Rl2), level of crowdedness and density
in
the trains (As1), journey time (Rl1), access to metro stations
(Rs4),
and security at metro stations (Rl3). The waiting time is
determined
as the most important attribute as we were expected and, also,
is
determined one of the three inDellOl et al. (2011)and one of
the
four important attributes inCelik et al. (2013). Another
attribute we
were expecting to be one of the most important attribute is
the
crowdedness, because more than 40% of the Istanbulers prefe
public transportation and it is usually crowded. Foote
(2004)note
that an improvement in crowdedness level increases CS by 7%
Therefore, we can conclude that crowdedness level is important
for
the passenger and increases CS if PT vehicles (train, bus etc.)
are lesscrowded. Also, journey time was one of the attribute that
we were
expecting to be one of top ve among rail transit users. Because
it is
one of the main reasons they prefer rail transit line instead of
other
type of PT. Accessing to rail transit lines` stations is more
difcul
than accessing to other types of PT stations, i.e., bus, jitney.
Since rai
transit lines' stations are mostly underground and take long
time to
access. Also we can inference the same result fromTable 2(time
to
trainrows 2329, columns 68). Time to train takes more than
10 min for 43% of the customers.Furthermore, based on the
results, the least important attribute
are up-to-datedness of the IUAS website (Rs5), service provided
by
IUAS phone (Rs6), costliness of interchanges (Tn8),
announcement
in stations during and after breakdowns (Tn6), and
announcement
in trains during and after breakdowns (Tn7). The averages of
theweights regarding all attributes are shown in Table 5.
3.5. Customer satisfaction evaluation
The averages of type-2 fuzzy performance values for RTN ve-
hicles are calculated using Eq.(2)in Step 2. The linguistic
terms are
converted into type-2 fuzzy numbers using six different scales
as
shown inTable 6by using Eq.(2). For instance, type-2 fuzzy
per-
formance value of M1 with respect to As1is evaluated
considering
frequencies for each linguistic scale, i.e., poor (24), medium
poor
(73), medium (75), medium good (98), good (640), and very
good
(166), and then the type-2 fuzzy performance ( eA M1s1 ) is
computed
as follows:
=
=
e
24 ((0; 1; 1; 3; 1; 1), (0. 5; 1; 1; 2; 0. 9; 0. 9))
73 ((1; 3; 3; 5; 1; 1), (2; 3; 3; 4; 0. 9; 0. 9))
75 ((3; 5; 5; 7; 1; 1), (4; 5; 5; 6; 0. 9; 0. 9))
98 ((5; 7; 7; 9; 1; 1), (6; 7; 7; 8; 0. 9; 0. 9))
640 ((7; 9; 9; 10; 1; 1), (8; 9; 9; 9. 5; 0. 9; 0. 9))
166 ((9; 10; 10; 10; 1; 1), (9. 5; 10; 10; 10; 0. 9; 0. 9))
/1076
((4.07; 5. 87; 5. 87; 7. 43; 1; 1),
(4. 97; 5. 87; 5. 87; 6. 65; 0. 9; 0. 9))
As M1 1
Similarly, the type-2 fuzzy performance values for each
criterionwith respect to each rail transit line are obtained, and
the results
are presented inTable 7.Then, the weighted type-2 fuzzy
performance values of the
RTN are calculated by multiplying the importance weights o
Table 1
Characteristics ofve rail transit lines.
Li ne Operating hours Li ne length
(km)
Daily
ridership
Trip time
(mins)
Number of
stations
Number of
cars
Number of daily
tripsaFrequency (min)b
Name Color
F1 Carroty 06:1524:00 0.6 30.000 2.5 2 4 195 3
M1 Red 06:0024:00 19.6 220.000 32 18 85 180 5M2 Green 06:1524:00
16.5 230.000 27 13 124 225 4
T1 Navy 06:0024:00 18.5 320.000 65 31 92 295 2T4 Orange
06:0024:00 15.3 95.000 42 22 78 165 5
a Single direction.b Frequency is based on the peak hours.
E. Celik et al. / Transport Policy 36 (2014) 283293 287
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
6/11
attributes (inTable 5) with type-2 fuzzy performance values
(in
Table 7), as described in Step 3.The positive ( *Pe ), negative
( Ne ) and weighted ( *Pv ) type-2
fuzzy ideal solutions for upper and lower reference points
are
determined using formulations in Step 4. Then, upper (SijU),
lower
(SijL) and average (Sij) group scores are calculated using Eq.
(4) in
Step 4, and the scores are shown in Table 8.In Step 5, nal
rankings based on averages and the worst group
scores are calculated using Eq. (6). Maximum group utility (v)
isconsidered as .5. Final rankings, and related regret and
average
scores are presented in Table 9. The smaller Q values
represent
higher CS level comparing other rail transit lines.Finally,
acceptable advantage, the Condition (1) in Step 5, be-
tween line F1, ( =Q 0.00F1 ) and M1 ( =Q 0.56F1 ), is
satised.Therefore, F1 has the best and M1 has the second best CS
scores.
RTN can be juxtaposed as F1, M1, M2, T4 and T1 from the best
to
the worst CS score, based on the survey.
3.6. Sensitivity analysis
In this subsection of the study, the concept of sensitivity
ana-
lysis investigates the impact of attributes with the proposed
in-
terval type-2 fuzzy VIKOR approach to validate the results of
CS
level to be steadier. The maximum group utility (v) is used
to
examine the ranking of RTN. This study assume that the v value
is
= 1while theQvalues of each alternative M1, M2, T1, T4, and
F1
are0, 32, 0.31, 1.00, 0.43and 0.00, respectively. The ranking
or-
der of the ve RTN is F14M24M14T44T1. When v value
is = 0.5, then the Q values of each RTN, M1, M2, T1, T4, and
F1,
are0, 56, 0.63, 0.88, 0.71and 0.00, respectively. The ranking
or-
der of the ve RTN is F14M14M24T44T1. If v value is = 0,
then the ranking order is F14T14M14M24T4. The Qvalue ofeach RTN,
M1, M2, T1, T4, and F1, is 0, 81, 0.94, 0.76, 1.00
and0.00, respectively.According to the aforementioned sample,
this study uses each
maximum group utility value, v , from 0.00 to 1.00 increasing
by
0.1 to examine the proposed approach, and then the obtained
results are found to be satisfactory, as shown in Table 10
and
graphically inFig. 2.The results show that, the variations of
thev values for each rail
transit lines changes rankings of the rail transit lines as in
Fig. 3.
The line F1 has the best CS rankings in all case-
s; =v 0.1, 0.2, .. , 1.0. T1 has the second best ranking
when
=v 0.0, but gets the worst score when v increases to1.0.
Fur-
thermore M1, M2, and T4's rankings improve as thev
increases.
M2 has the second best ranking at almost all cases.
Table 2
Demographic and transportation information (n4966).
Question Option Frequency Percent Question Option Frequency
Percent
Survey Date Monday 629 12.7 Access (by) Walk 2809 56.6Tuesday
780 15.7 Tram 563 11.3
Wednesday 710 14.3 Bus 805 16.2
Thursday 836 16.8 Cab 52 1.0
Friday 835 16.8 Shuttle 25 0.5
Saturday 676 13.6 Private car 146 2.9Sunday 500 10.1 Jitney 188
3.8
Survey Time Morning (peak hour) 807 16.9 Sea bus 125 2.5
Morning 667 14.0 . Other 253 5.1
Noon 1641 34.4 Disabled Yes 77 1.6
Evening (peak hour) 1175 24.6 No 4889 98.4Evening 477 10.0 Has
private car Yes 1995 40.2
Age 1425 2317 46.7 No 2971 59.8
2635 1312 26.4 Aim for use Homework 2225 44.8
3645 675 13.6 Homeschool 1037 20.9
4655 397 8.0 Shopping 156 3.1
56 265 5.3 Business 661 13.3Gender Male 3844 77.4 Social 602
12.1
Female 1122 22.6 Medical services 76 1.5Education Non-educated
32 0.6 Visit 201 4.0
Primary school (5 years) 568 11.5 Other 8 0.2Middle school (8
years) 493 9.9 Time to train 010 2839 57.2
High School (11 years) 1711 34.5 (min) 11
20 1128 22.7Associate degree 123 2.5 2130 453
9.1Undergraduate(stud) 950 19.2 3140 153 3.1
Bachelor of Sc. 920 18.6 4150 152 3.1
Graduate (M.S., Ph.D.) 162 3.3 5160 130 2.6Has a job Yes 3192
64.3 60 111 2.2
No 1774 35.7 Travel time 010 399 8.0Marital status Married 1862
37.5 (min) 1120 1319 26.6
Single 3104 62.5 2130 1148 23.1Income NA 126 2.5 3140 661
13.3(TL/month) 01000 748 15.1 4150 509 10.2
10012000 2122 42.7 5160 352 7.1
20013000 1129 22.7 60 578 11.630014000 367 7.4 Day of Use
Weekday 2908 58.6
4001-5000 248 5.0 Saturday 134 2.7
5000 226 4.6 Sunday 117 2.4
Ticket kind Token 475 9.6 Weekday and Saturday 638 12.8
Full Fare 2605 52.5 Weekday and Sunday 30 0.6
Student fare 1719 34.6 Saturday and Sunday 252 5.1
Discount fare 167 3.4 All week 887 17.9
E. Celik et al. / Transport Policy 36 (2014) 283293288
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
7/11
This study conrms that the results of the ranking orders of
all
ve rail transit lines, by using the proposed approach, are
con-
sistent. Furthermore, the proposed approach nds the gap be-
tween the Q values of various rail transit lines. Q values get
smaller
when the maximum group utility value is increased from 0.1
to
1.0. According to the analysis above, this paper nds that
the
proposed approach produces satisfactory results and
providesproper information to assist managers in decision
making.
4. Discussion and conclusion
Railway transportation is one of the most important PT types
especially in big and crowded cities, i.e., Istanbul. In such
cities,
RTN is used as the rst way to escape from trafc congestion,
specically during the rush hours. Therefore, getting a high
CS
level is very important for municipalities and governments.
In
order to assess the CS level, IUAS conducted a survey
considering
ve different rail transit lines in Istanbul. The satisfaction
level of
the Istanbulers related to RTN is questioned in this survey.
Hence, we present a novel approach which integrates
statistica
analysis, SERVQUAL, type-2 fuzzy sets and VIKOR to evaluate
CS
level for the RTN in Istanbul. The attributes which have the
highest
and the lowest CS scores are determined for each rail transit
lines
according to the results obtained by proposed approach.Opricovic
(1998)and Opricovic and Tzeng (2004) suggest that
the weight of the strategy of maximum group utility (v)
value
should generally be taken0.5. The rankings are occurred as
F14M14M24T44T1 when v is considered as0.5. This meansrail
transit line F1 rated as the best line by passengers
considering
all 26 attributes together. TheQ(minimum) values are
determined
as 0.00, 0.56, 0.63, 0.71and 0.88 for F1, M1, M2, T4, and T1
re-
spectively as shown inTable 10row 7.Furthermore, the PT
providers frequently states the comfort
attribute by favoring enhanced standards for cars or stations.
In
their study, Wall and McDonald (2007), aimed to increase the
journey comfort by introducing of a new eet of buses. This
im-
provement is highlighted by the customers as one of the most
important effects in transporting by PT services.
Additionally
European Local Transport Information Service (2010) notes
that
providing new covered shelters at the rail stations in
Norwich/
England increased CS levels and the 98% of the respondents
were
satised by the quality. In our study, most of the respondents
are
Table 3
Dimensions and attributes of CS evaluation for rail transit
services.
Dimension Attribute
Assurance Level of crowdedness and density in the trains
(As1)Noise level and vibration during the journey (As2)
Lighting in the stations (As3)
Air-conditioning system of trains interior (temperature, hu-
midity) (As4)
Escalators, elevators and bent conveyors (As5)Comfort level in
the stations (As6)
Empathy Attitude and behaviors of the security staff (Em1)
Reliability Journey time (Rl1)Waiting time before departure
(Rl2)
Security at stations (Rl3)
Security inside trains (Rl4)
Arrival performance with respect to schedules (Rl5)
Responsiveness Ticketing service (Rs1)
Ticket vending machines/services (Rs2)
Smooth functioning of the turnstiles (Rs3)
Access to stations (Rs4)
Up-to-datedness of the IUAS website (Rs5)Service provided by
IUAS phone (Rs6)
Tangibles Costliness of ticket (Tn1)Usage of modern equipment in
stations (Screen, schedule,
routes) (Tn2)
Cleanliness of stations (Tn3)
Cleanliness of train interior (Tn4)
Usage of modern equipment inside the trains services
(Screens, route map, announcement) (Tn5)
Announcement in stations during and after breakdowns
(Tn6)
Announcement in trains during and after breakdowns (Tn7)
Costliness of interchanges (Tn8)
Table 4
Linguistic terms of the weights of attributes.
Linguistic terms Interval type-2 fuzzy sets
Medium low ((0.1;0.3;0.3;0.5;1;1),
(0.2;0.3;0.3;0.4;0.9;0.9))
Medium ( (0.3;0.5;0.5;0.7; 1; 1) , (0.4;0.5;0.5; 0. 6;0.
9;0.9))
Medium high ((0.5;0.7;0.7;0.9;1;1),
(0.6;0.7;0.7;0.8;0.9;0.9))High ((0.7;0.9;0.9;1;1;1),
(0.8;0.9;0.9;0.95;0.9;0.9))
Very hig h ( (0.9;1;1;1;1;1), (0.95;1;1;1;0.9;0.9) )
Table 5
The importance weights of the attributes.
Attributes Weights
As1 ((0.618;0.784;0.784;0.896;1;1),
(0.701;0.784;0.784;0.84;0.9;0.9))As2
((0.406;0.599;0.599;0.768;1;1),
(0.503;0.599;0.599;0.684;0.9;0.9))
As3 ((0.366;0.56;0.56;0.734;1;1),
(0.463;0.56;0.56;0.647;0.9;0.9))As4 ((0.396;0.587;0.587;0.755;1;1),
(0.492;0.587;0.587;0.671;0.9;0.9))As5 ((0.325;0.
517;0.517;0.693;1;1), (0.421;0.517;0.517;0.605;0.9;0.9))
As6 ((0.315;0.508;0.508;0.686;1; 1),
(0.411;0.508;0.508;0.597;0.9;0.9))
Em1 ((0.519;0.7;0.7;0.839;1;1), (0.609;0.7;
0.7;0.769;0.9;0.9))
Rl1 ((0.57;0.747;0.747;0.875;1;1),
(0.658;0.747;0.747;0.811;0.9;0.9))Rl2
((0.689;0.835;0.835;0.911;1;1),
(0.762;0.835;0.835;0.873;0.9;0.9))
Rl3 ((0.586;0. 767;0.767;0.89;1;1), (0.677;0.767;0.
767;0.828;0.9; 0.9))Rl4 ((0.53;0.716;0.716;0.856; 1;1),
(0.623;0.716;0.716;0.786;0.9;0.9))Rl5
((0.421;0.612;0.612;0.777;1;1),
(0.517;0.612;0.612;0.695;0.9;0.9))
Rs1 ((0.43;0.619;0.619;0.781; 1;1), (0.525;0.619;0.619;0,7;0.
9;0.9))Rs2 ((0.343;0.538;0.538;0.716;1;1),
(0.441;0.538;0.538;0.627;0.9;0.9))Rs3
((0.295;0.491;0.491;0.674;1;1),
(0.393;0.491;0.491;0.583;0.9;0.9))Rs4 ((0.565;0.
741;0.741;0.863;1;1), (0.653;0.741;0.741;0.802;0.9;0.9))Rs5
((0.336;0. 518;0.518;0.673;1;1),
(0.427;0.518;0.518;0.595;0.9;0.9))Rs6 ((0.1;0.3;0.3;0.5;1;1),
(0.2;0.3;0.3;0.4;0.9;0.9))
Tn1 ((0.516;0.693;0.693;0.829;1;1),
(0.605;0.693;0.693;0.761;0.9;0.9))
Tn2 ((0.388;0.581;0.581;0.752;1;1),
(0.485;0.581;0.581;0.667;0.9;0.9))
Tn3 ((0.471;0.659;0.659;0.816;1;1),
(0.565;0.659;0.659;0.737;0.9;0.9))Tn4
((0.42;0.611;0.611;0.777;1;1), (0.516;0.611;0.611;0.694;0.9;0.
9))Tn5 ((0.299;0.494;0.494;0.676;1;1),
(0.397;0.494;0.494;0.585;0.9;0.9))Tn6
((0.352;0.543;0.543;0.715;1;1),
(0.447;0.543;0.543;0.629;0.9;0.9))Tn7
((0.294;0.487;0.487;0.666;1;1),
(0.39;0.487;0.487;0.576;0.9;0.9))Tn8
((0.378;0.565;0.565;0.727;1;1),
(0.472;0.565;0.565;0.646;0.9;0.9))
Table 6
Linguistic terms for rail transit line rating.
Linguistic terms Interval type-2 fuzzy sets
Poor ((0;1;1;3;1;1), (0.5;1;1;2;0.9;0.9))
Medium poor ((1;3;3;5;1;1), (2;3;3;4;0.9;0.9))
Medium ((3;5;5;7;1;1), (4;5;5;6;0.9;0.9))Medium good
((5;7;7;9;1;1), (6;7;7;8;0.9;0.9))Good ((7;9;9;10;1;1),
(8;9;9;9.5;0.9;0.9))
Very good ((9;10;10;10;1;1), (9.5;10;10;10;0.9;0.9))
E. Celik et al. / Transport Policy 36 (2014) 283293 289
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
8/11
Table 7
The type-2 fuzzy performance values.
Attributes M1 M2 T1
As1
((4.07;5.87;5.87;7.43;1;1),(4.97;5.87;5.87;6.65;0.9;0.9))
((4.96;6.8;6.8;8.24;1;1),(5.88;6.8;6.8;7.52;0.9;0.9))
((2.15;3.78;3.78;5.64;1;1),(2.97;3.78;3.78;4.71;0.9;0.9))
As2 ((5.69;7.52;7.52;8.77;1;1),
(6.6;7.52;7.52;8.15;0.9;0.9))
((6.42;8.3;8.3;9.46;1;1),
(7.36;8.3;8.3;8.88;0.9;0.9))
((5.05;6.95;6.95;8.43;1;1),
(6;6.95;6.95;7.69;0.9;0.9))As3 ((7;8.81;8.81;9.73;1;1),
(7.9;8.81;8.81;9.27;0.9;0.9))
((6.87;8.72;8.72;9.74;1;1),
(7.8;8.72;8.72;9.23;0.9;0.9))
((6.39;8.29;8.29;9.45;1;1),
(7.34;8.29;8.29;8.87;0.9;0.9))As4
((5.79;7.62;7.62;8.84;1;1),
(6.71;7.62;7.62;8.23;0.9;0.9))
((5.94;7.8;7.8;9.04;1;1),
(6.87;7.8;7.8;8.42;0.9;0.9))
((4.55;6.37;6.37;7.9;1;1),
(5.46;6.37;6.37;7.14;0.9;0.9))As5
((6.72;8.53;8.53;9.53;1;1),
(7.63;8.53;8.53;9.03;0.9;0.9))
((6.37;8.23;8.23;9.38;1;1),
(7.3;8.23;8.23;8.81;0.9;0.9))
((5.86;7.75;7.75;9.03;1;1),
(6.81;7.75;7.75;8.39;0.9;0.9))As6 ((6.46;8.31;8.31;9.4;1;1),
(7.39;8.31;8.31;8.85;0.9;0.9))((6.61;8.49;8.49;9.59;1;1),(7.55;8.49;8.49;9.04;0.9;0.9))
((5.74;7.66;7.66;8.99;1;1),(6.7;7.66;7.66;8.32;0.9;0.9))
Em1 ((6.67;8.5;8.5;9.54;1;1),
(7.59;8.5;8.5;9.02;0.9;0.9))
((6.52;8.34;8.34;9.43;1;1),
(7.43;8.34;8.34;8.88;0.9;0.9))
((6.61;8.41;8.41;9.45;1;1),
(7.51;8.41;8.41;8.93;0.9;0.9))
Rl1 ((6.91;8.69;8.69;9.62;1;1),
(7.8;8.69;8.69;9.16;0.9;0.9))
((6.92;8.72;8.72;9.7;1;1),
(7.82;8.72;8.72;9.21;0.9;0.9))
((5.77;7.68;7.68;9.02;1;1),
(6.73;7.68;7.68;8.35;0.9;0.9))Rl2 ((6.28;8.11;8.11;9.2;1;1),
(7.2;8.11;8.11;8.66;0.9;0.9))
((6.66;8.47;8.47;9.52;1;1),
(7.56;8.47;8.47;9;0.9;0.9))
((5.64;7.55;7.55;8.9;1;1),
(6.6;7.55;7.55;8.22;0.9;0.9))Rl3 ((6.53;8.37;8.37;9.42;1;1),
(7.45;8.37;8.37;8.89;0.9;0.9))
((6.42;8.26;8.26;9.38;1;1),
(7.34;8.26;8.26;8.82;0.9;0.9))
((6.19;8.07;8.07;9.25;1;1),
(7.13;8.07;8.07;8.66;0.9;0.9))Rl4 ((6.38;8.21;8.21;9.3;1;1)
(7.29;8.21;8.21;8.76;0.9;0.9))
((6.53;8.37;8.37;9.47;1;1),
(7.45;8.37;8.37;8.92;0.9;0.9))
((6.05;7.92;7.92;9.13;1;1),
(6.99;7.92;7.92;8.53;0.9;0.9))Rl5 ((6.7;8.54;8.54;9.56;1;1),
(7.62;8.54;8.54;9.05;0.9;0.9))
((6.76;8.62;8.62;9.66;1;1),
(7.69;8.62;8.62;9.14;0.9;0.9))
((6.09;8;8;9.26;1;1), (7.05;8;8;8.63;0.9;0.9))
Rs1 ((6.67;8.48;8.48;9.47;1;1),
(7.57;8.48;8.48;8.97;0.9;0.9))
((6.47;8.32;8.32;9.42;1;1),
(7.39;8.32;8.32;8.87;0.9;0.9))
((6.22;8.08;8.08;9.24;1;1),
(7.15;8.08;8.08;8.66;0.9;0.9))Rs2
((6.79;8.59;8.59;9.56;1;1),
(7.69;8.59;8.59;9.08;0.9;0.9))
((6.62;8.48;8.48;9.57;1;1),
(7.55;8.48;8.48;9.02;0.9;0.9))
((6.36;8.24;8.24;9.4;1;1)
(7.3;8.24;8.24;8.82;0.9;0.9))Rs3 ((7.1;8.91;8.91;9.81;1;1),
(8;8.91;8.91;9.36;0.9;0.9))
((6.85;8.7;8.7;9.72;1;1),
(7.77;8.7;8.7;9.21;0.9;0.9))
((6.68;8.55;8.55;9.63;1;1),
(7.61;8.55;8.55;9.09;0.9;0.9))Rs4 ((6.66;8.5;8.5;9.53;1;1),
(7.58;8.5;8.5;9.02;0.9;0.9))
((6.6;8.45;8.45;9.54;1;1),
(7.53;8.45;8.45;8.99;0.9;0.9))
((6.24;8.15;8.15;9.35;1;1),
(7.19;8.15;8.15;8.75;0.9;0.9))Rs5
((6.05;7.81;7.81;8.95;1;1),
(6.93;7.81;7.81;8.38;0.9;0.9))
((6.57;8.03;8.03;8.96;1;1),
(7.3;8.03;8.03;8.5;0.9;0.9))
((5.92;7.81;7.81;8.95;1;1),
(6.93;7.81;7.81;8.38;0.9;0.9))Rs6
((5.14;6.82;6.82;8.05;1;1),
(5,98;6,82;6,82;7,44;0,9;0,9))
((6.46;7.88;7.88;8.79;1;1),
(7.17;7.88;7.88;8.34;0.9;0.9))
((5.52;6.82;6.82;8.05;1;1),
(5.98;6.82;6.82;7.44;0.9;0.9))
Tn1 ((4.51;6.24;6.24;7.65;1;1),
(5.38;6.24;6.24;6.94;0.9;0.9))
((5.32;7.13;7.13;8.47;1;1),
(6.22;7.13;7.13;7.8;0.9;0.9))
((4.2;6.02;6.02;7.61;1;1),
(5.11;6.02;6.02;6.82;0.9;0.9))Tn2
((6.85;8.68;8.68;9.65;1;1),(7.77;8.68;8.68;9.16;0.9;0.9))
((6.73;8.59;8.59;9.64;1;1),(7.66;8.59;8.59;9.11;0.9;0.9))
((6.41;8.29;8.29;9.44;1;1),(7.35;8.29;8.29;8.86;0.9;0.9))
Tn3 ((6.66;8.47;8.47;9.49;1;1),
(7.57;8.47;8.47;8.98;0.9;0.9))
((6.64;8.5;8.5;9.58;1;1),
(7.57;8.5;8.5;9.04;0.9;0.9))
((5.98;7.88;7.88;9.14;1;1),
(6.93;7.88;7.88;8.51;0.9;0.9))
Tn4 ((6.56;8.35;8.35;9.37;1;1),
(7.45;8.35;8.35;8.86;0.9;0.9))
((6.63;8.48;8.48;9.57;1;1),
(7.55;8.48;8.48;9.03;0.9;0.9))
((6.17;8.06;8.06;9.28;1;1),
(7.11;8.06;8.06;8.67;0.9;0.9))Tn5 ((6.94;8.75;8.75;9.7;1;1),
(7.84;8.75;8.75;9.22;0.9;0.9))
((6.73;8.58;8.58;9.65;1;1),
(7.66;8.58;8.58;9.12;0.9;0.9))
((6.46;8.34;8.34;9.47;1;1),
(7.4;8.34;8.34;8.9;0.9;0.9))Tn6 ((6.92;8.73;8.73;9.68;1;1),
(7.82;8.73;8.73;9.2;0.9;0.9))
((6.62;8.49;8.49;9.59;1;1),
(7.55;8.49;8.49;9.04;0.9;0.9))
((6.1;7.97;7.97;9.18;1;1),
(7.03;7.97;7.97;8.58;0.9;0.9))Tn7
((6.99;8.79;8.79;9.72;1;1),
(7.89;8.79;8.79;9.26;0.9;0.9))
((6.58;8.44;8.44;9.55;1;1),
(7.51;8.44;8.44;8.99;0.9;0.9))
((6.3;8.17;8.17;9.34;1;1),
(7.24;8.17;8.17;8.75;0.9;0.9))Tn8 ((4.95;6.7;6.7;8.02;1;1),
(5.82;6.7;6.7;7.36;0.9;0.9))
((6.52;7.93;7.93;8.84;1;1),
(7.23;7.93;7.93;8.39;0.9;0.9))
((5.19;6.7;6.7;8.02;1;1),
(5.82;6.7;6.7;7.36;0.9;0.9))Attributes T4 F1
As1 ((5.08;6.92;6.92;8.32;1;1), (6;6.92;6.92;7.62;0.9;0.9))
((4.27;6.06;6.06;7.64;1;1), (5.16;6.06;6.06;6.85;0.9;0.9))
As2 ((6.03;7.88;7.88;9.12;1;1), (6.96;7.88;7.88;8.5;0.9;0.9))
((6.25;8.05;8.05;9.2;1;1), (7.15;8.05;8.05;8.62;0.9;0.9))As3
((6.7;8.55;8.55;9.64;1;1), (7.63;8.55;8.55;9.1;0.9;0.9))
((7.32;8.98;8.98;9.75;1;1), (8.15;8.98;8.98;9.37;0.9;0.9))
As4 ((6.05;7.89;7.89;9.09;1;1), (6.97;7.89;7.89;8.49;0.9;0.9))
((5.9;7.66;7.66;8.82;1;1), (6.78;7.66;7.66;8.24;0.9;0.9))As5
((6.58;8.43;8.43;9.54;1;1), (7.5;8.43;8.43;8.98;0.9;0.9))
((6.93;8.64;8.64;9.55;1;1), (7.79;8.64;8.64;9.1;0.9;0.9))As6
((6.46;8.32;8.32;9.48;1;1), (7.39;8.32;8.32;8.9;0.9;0.9))
((6.78;8.54;8.54;9.54;1;1), (7.66;8.54;8.54;9.04;0.9;0.9))
Em1 ((6.51;8.35;8.35;9.46;1;1), (7.43;8.35;8.35;8.91;0.9;0.9))
((7.23;8.81;8.81;9.59;1;1), (8.02;8.81;8.81;9.2;0.9;0.9))
Rl1 ((6.36;8.2;8.2;9.34;1;1), (7.28;8.2;8.2;8.77;0.9;0.9))
((7.29;8.9;8.9;9.69;1;1), (8.1;8.9;8.9;9.3;0.9;0.9))Rl2
((6.19;8.05;8.05;9.28;1;1), (7.12;8.05;8.05;8.67;0.9;0.9))
((7.24;8.89;8.89;9.72;1;1), (8.07;8.89;8.89;9.31;0.9;0.9))Rl3
((6.49;8.33;8.33;9.45;1;1), (7.41;8.33;8.33;8.89;0.9;0.9))
((6.84;8.52;8.52;9.44;1;1), (7.68;8.52;8.52;8.98;0.9;0.9))
Rl4 ((6.47;8.32;8.32;9.44;1;1), (7.4;8.32;8.32;8.88;0.9;0.9))
((6.94;8.59;8.59;9.47;1;1), (7.77;8.59;8.59;9.03;0.9;0.9))Rl5
((6.62;8.48;8.48;9.58;1;1), (7.55;8.48;8.48;9.03;0.9;0.9))
((7.32;8.91;8.91;9.67;1;1), (8.12;8.91;8.91;9.29;0.9;0.9))Rs1
((6.47;8.31;8.31;9.42;1;1), (7.39;8.31;8.31;8.86;0.9;0.9))
((6.73;8.4;8.4;9.32;1;1), (7.57;8.4;8.4;8.86;0.9;0.9))
Rs2 ((6.61;8.46;8.46;9.56;1;1), (7.54;8.46;8.46;9.01;0.9;0.9))
((6.89;8.61;8.61;9.55;1;1), (7.75;8.61;8.61;9.08;0.9;0.9))Rs3
((6.69;8.55;8.55;9.64;1;1), (7.62;8.55;8.55;9.1;0.9;0.9))
((5.9;7.58;7.58;8.96;1;1), (6.74;7.58;7.58;8.27;0.9;0.9))Rs4
((6.63;8.47;8.47;9.55;1;1), (7.55;8.47;8.47;9.01;0.9;0.9))
((7.26;8.89;8.89;9.68;1;1), (8.08;8.89;8.89;9.29;0.9;0.9))
Rs5
((5.81;7.66;7.66;8.96;1;1).(6.74;7.66;7.66;8.31;0.9;0.9))
((6.38;8.23;8.23;9.38;1;1).(7.3;8.23;8.23;8.8;0.9;0.9))
E. Celik et al. / Transport Policy 36 (2014) 283293290
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
9/11
happy with the quality of the lighting system in the trains (in
the
lines F1, M1, M2, and T4) and the provided comfort level both
in
the train and stations. The comfort level provided by the PT
service
supplier may also increases the number of ridership as noted
in
Foote (2004). Foote (2004) analyzed the improvements that
arerelated to comfort level issues, i.e., safety, cleanliness, in
the Chi-
cago Transit Authority's PT services and they showed that
the
number of trips is increased by 5% (per annum) after the
comfort
level is improved. Besides lighting, passengers are happy and
very
satised with the smooth functioning of turnstiles, usage of
modern equipment in stations and trains and arrival
performances
with respect to schedules, and journey time.Speed affects
journey time and the arrival performance of the
trains. Therefore, speed is also critical variable in affecting
CS of PT
Table 8
Upper, lower and average group scores.
M1 M2 T1 T4 F1
Attributes
S S,ij
UijL Sij
S S,ij
UijL Sij
S S,ij
UijL Sij
S S,ij
UijL Sij
S S,ij
UijL Sij
As1 [0.81;0.52] 0.67 [0.8;0.48] 0.64 [0.96;0.85] 0.90 [0.8;0.47]
0.64 [0.81;0.51] 0.66As2 [1.24;0.98] 1.11 [1.26;0.96] 1.11
[1.26;1.04] 1.15 [1.25;0.96] 1.11 [1.25;0.95] 1.10
As3 [1.56;1.39] 1.47 [1.57;1.4] 1.48 [1.57;1.42] 1.50 [1.57;1.4]
1.49 [1.55;1.37] 1.46As4 [1.17;0.86] 1.02 [1.18;0.87] 1.02
[1.18;0.95] 1.07 [1.18;0.87] 1.02 [1.16;0.85] 1.01As5 [1.41;1.18]
1.29 [1.41;1.19] 1.30 [1.41;1.22] 1.31 [1.41;1.19] 1.30 [1.4;1.17]
1.29
As6 [1.36;1.17] 1.26 [1.37;1.17] 1.27 [1.36;1.2] 1.28
[1.37;1.17] 1.27 [1.36;1.16] 1.26
Em1 [1.63;1.54] 1.59 [1.63;1.56] 1.60 [1.62;1.54] 1.58
[1.64;1.56] 1.60 [1.58;1.49] 1.54
Rl1 [1.35;1.03] 1.19 [1.36;1.03] 1.20 [1.41;1.18] 1.30
[1.37;1.08] 1.23 [1.32;1] 1.16Rl2 [1.2;0.95] 1.07 [1.18;0.89] 1.04
[1.27;1.1] 1.18 [1.22;0.97] 1.09 [1.14;0.84] 0.99Rl3 [1.54;1.48]
1.51 [1.55;1.49] 1.52 [1.56;1.51] 1.54 [1.55;1.49] 1.52 [1.51;1.45]
1.48Rl4 [1.54;1.39] 1.46 [1.54;1.38] 1.46 [1.55;1.43] 1.49
[1.54;1.39] 1.47 [1.5;1.35] 1.43
Rl5 [1.5;1.25] 1.37 [1.51;1.25] 1.38 [1.52;1.3] 1.41 [1.51;1.26]
1.38 [1.47;1.21] 1.34
Rs1 [1.54;1.49] 1.52 [1.55;1.5] 1.52 [1.55;1.5] 1.53 [1.55;1.5]
1.52 [1.52;1.47] 1.49Rs2 [1.55;1.51] 1.53 [1.56;1.52] 1.54
[1.56;1.53] 1.54 [1.56;1.52] 1.54 [1.54;1.5] 1.52Rs3 [1.36;1.01]
1.18 [1.36;1.01] 1.19 [1.36;1.01] 1.19 [1.36;1.02] 1.19 [1.34;1.05]
1.20Rs4 [1.48;1.31] 1.40 [1.49;1.32] 1.41 [1.51;1.37] 1.44
[1.49;1.32] 1.40 [1.44;1.26] 1.35Rs5 [1.31;1.22] 1.27 [1.29;1.18]
1.23 [1.32;1.22] 1.27 [1.33;1.24] 1.29 [1.34;1.23] 1.29Rs6
[0.96;0.64] 0.80 [0.99;0.64] 0.81 [0.96;0.64] 0.80 [0.98;0.64] 0.81
[0.95;0.66] 0.80
Tn1 [1.17;0.99] 1.08 [1.18;0.93] 1.05 [1.2;1.03] 1.12
[1.18;0.93] 1.05 [1.17;1] 1.09Tn2 [1.55;1.51] 1.53 [1.56;1.52] 1.54
[1.56;1.53] 1.54 [1.56;1.52] 1.54 [1.54;1.5] 1.52
Tn3 [1.49;1.29] 1.39 [1.5;1.3] 1.40 [1.51;1.35] 1.43 [1.5;1.3]
1.40 [1.47;1.28] 1.37Tn4 [1.52;1.43] 1.47 [1.54;1.44] 1.49
[1.54;1.46] 1.50 [1.53;1.44] 1.49 [1.51;1.42] 1.47Tn5 [1.53;1.48]
1.50 [1.53;1.48] 1.51 [1.53;1.48] 1.51 [1.53;1.48] 1.51 [1.51;1.46]
1.49Tn6 [1.45;1.28] 1.36 [1.46;1.28] 1.37 [1.45;1.3] 1.38
[1.46;1.28] 1.37 [1.44;1.27] 1.35Tn7 [1.47;1.35] 1.41 [1.48;1.35]
1.42 [1.47;1.36] 1.42 [1.48;1.35] 1.41 [1.46;1.34] 1.40Tn8
[1.27;1.01] 1.14 [1.24;0.91] 1.08 [1.26;1.01] 1.13 [1.26;0.94] 1.10
[1.27;1.01] 1.14
Table 9
The nal rankings for ve rail transit lines.
M1 M2 T1 T4 F1
Si 33.60 33.58 34.48 33.74 33.18
Ri 1.59 1.60 1.58 1.60 1.54
=Q v( 0.5)i 0.56 0.63 0.88 0.71 0.00
Table 10
TheQi values for different maximum group utilities.
M1 M2 T1 T4 F1
v0.0 0.81 0.94 0.76 1.00 0.00
v0.1 0.76 0.88 0.78 0.94 0.00
v0.2 0.71 0.81 0.81 0.89 0.00
v0.3 0.66 0.75 0.83 0.83 0.00
v0.4 0.61 0.69 0.86 0.77 0.00
v0.5 0.56 0.63 0.88 0.71 0.00
v0.6 0.52 0.56 0.90 0.66 0.00
v0.7 0.47 0.50 0.93 0.60 0.00
v0.8 0.42 0.44 0.95 0.54 0.00
v0.9 0.37 0.37 0.98 0.48 0.00
v1.0 0.32 0.31 1.00 0.43 0.00
Table 7 (continued )
Attributes T4 F1
Rs6 ((5.25;7.08;7.08;8,33;1;1), (6,17;7,08;7,08;7,71;0,9;0,9))
((4,25;6;6;7,5;1;1), (5,13;6;6;6,75;0,9;0,9))
Tn1 ((5,44;7,25;7,25;8,56;1;1), (6,35;7,25;7,25;7,9;0,9;0,9))
((4,44;6,12;6,12;7,56;1;1), (5,28;6,12;6,12;6,84;0,9;0,9))Tn2
((6,69;8,54;8,54;9,62;1;1), (7,62;8,54;8,54;9,08;0,9;0,9))
((6.8;8.53;8.53;9.49;1;1).(7.67;8.53;8.53;9.01;0.9;0.9))Tn3
((6.61;8.44;8.44;9.54;1;1).(7.52;8.44;8.44;8.99;0.9;0.9))
((6.98;8.67;8.67;9.56;1;1).(7.83;8.67;8.67;9.12;0.9;0.9))Tn4
((6.58;8.43;8.43;9.53;1;1).(7.5;8.43;8.43;8.98;0.9;0.9))
((6.87;8.59;8.59;9.53;1;1), (7.73;8.59;8.59;9.06;0.9;0.9))
Tn5 ((6.66;8.5;8.5;9.61;1;1), (7.58;8.5;8.5;9.06;0.9;0.9))
((6.89;8.63;8.63;9.56;1;1), (7.76;8.63;8.63;9.1;0.9;0.9))Tn6
((6.65;8.48;8.48;9.58;1;1), (7.56;8.48;8.48;9.03;0.9;0.9))
((6.75;8.49;8.49;9.47;1;1), (7.62;8.49;8.49;8.98;0.9;0.9))
Tn7 ((6.64;8.48;8.48;9.58;1;1), (7.56;8.48;8.48;9.03;0.9;0.9))
((6.7;8.43;8.43;9.4;1;1), (7.57;8.43;8.43;8.91;0.9;0.9))Tn8
((5.86;7.53;7.53;8.63;1;1), (6.7;7.53;7.53;8.08;0.9;0.9))
((5.02;6.77;6.77;8.11;1;1), (5.9;6.77;6.77;7.44;0.9;0.9))
E. Celik et al. / Transport Policy 36 (2014) 283293 291
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
10/11
services (Redman et al., 2013). US Federal Transit
Administration
(2010)notes that the ridership is increased by 24.5% over one
year
after the commuting time is reduced by 15 min each way
betweencentral New York stations and outlying areas.
On the other hand, there are several attributes need to be
im-
proved in order to increase the CS level for RTN in
Istanbul.
Crowdedness and density of passengers is determined as one
of
the attribute that needs to be improved. Adding new rail
transit
line, increasing number of cars per trip are the two suggestions
to
be made to reduce the crowdedness. Also, re-optimization of
the
schedules should be considered, to reduce the crowdedness,
considering the changes in the crowdedness level. Besides,
opti-
mization of the speed arrangement between stations is an
im-portant factor that affects the crowdedness. Pucher et al.
(2005)
suggest adding new bus lanes in Seoul, Korea aimed at
improving
the speed of PT. Thus, putting new rail transit line(s) into
services
will affect the speed and consequently the crowdedness level
atstations. There are several new rail transit lines under
construction
for Istanbulers (IUAS, 2013), i.e., M5Uskudar/Sancaktepe,
M6Levent/Rumelihisarustu, M7Mecidiyekoy/Mahmutbey, inFig. 1. A
subsequent survey should be conducted, once the new rail
transit
lines are putted into service, to analyze CS level. Moreover,
Air-
conditioning is determined as another attribute that need to
be
enhanced because air conditioning systems on vehicles are one
of
the motivations to use PT (Beiro and Sarseld Cabral, 2007).
Ac-
cording to IUAS, air-conditioning system fails during the days
andit is not possible to x it while the trains are on the move.
Therefore, preventive maintenance of the air-conditioning
system,
during the off times, is important to prevent fails. Another
attri-
bute that causes low CS level is noise and vibration during
the
journey. As in air-conditioning, preventive maintenance of
the
trains and rail transit lanes are the key subjects in reducing
noiseand vibration. One of the applied preventive maintenance
activity
is the periodic and continuous inspection of the rail lanes.
Re-duction in noise and vibration will provide a comfortable
journey
to the passengers. Lastly, the phone service is another
attribute
that resulted in low CS level. Training of the phone service
per-
sonnel and increasing the number of personnel should improve
CSlevel.
In summary, the attributes need to be improved are de-termined,
and, for all lines different improvement suggestions are
proposed. The contributions of the paper to the literature are
asfollows: (1) it proposes a novel CS evaluation approach for RTN
of
Istanbul, by using survey study, statistical analysis and MADM.
By
the integration of these three methods together, the CS levels
can
be analyzed and evaluated in a healthy manner; (2) an
integratednovel interval type-2 fuzzy MADM method is proposed based
on
SERVQUAL and VIKOR to evaluate and improve CS in Istanbul
RTN.Hence, the proposed MADM benets from the advantages of all
three methods. Also, the interval type-2 fuzzy sets and VIKOR
in-
tegration reveals and solves ambiguity and uncertainty in a
more
realistic way; (3) the proposed method provides directions for
the
future investments that can be made; (4) the proposed method
can be generalized and applied to complex decision making
pro-blems encountering inexact, indenite and subjective data or
un-certain information and (5) it is aimed that, the proposed
method
will be used in big and crowded cities' rail transit activities
like
Istanbul, to evaluate and improve CS levels by policy and
decision
makers.As aforementioned, IMM and IUAS have plans to quadruple
the
length of the total RTN of Istanbul by the end of 2019. As a
futuredirection, for new lines in Istanbul or for different
countries` RTN,
this study can be taken as a reference point in terms of CS
eva-luation and the determination of the most important and
vital
attributes to improve.
Acknowledgment
The authors would like to express their gratitude to IUAS
(Is-
tanbul Public Transportation Co.) for their understanding,
support,and the data provided. Finally, the authors would like to
thank the
two anonymous referees for their helpful comments and sug-
gested improvements.
References
Awasthi, A., Chauhan, S.S., Omrani, H., Panahi, A., 2011. A
hybrid approach based onSERVQUAL and fuzzy TOPSIS for evaluating
transportation service quality.Comput. Ind. Eng. 61 (3),
637646.
Beiro, G., Sarseld Cabral, J.A., 2007. Understanding attitudes
towards public
transport and private car: a qualitative study. Transp. Policy
14 (6), 478
489.Celik, E., Bilisik, O.N., Erdogan, M., Gumus, A.T., Baracli,
H., 2013. An integrated novelinterval type-2 fuzzy MCDM method to
improve customer satisfaction in publictransportation for Istanbul.
Transp. Res. Part E: Log. Transp. Rev. 58, 2851.
Chen, S.M., Lee, L.W., 2010. Fuzzy multiple attributes group
decision-making basedon the interval type-2 TOPSIS method. Expert
Syst. Appl. 37 (4), 27902798.
Chou, C.C., Liu, L.J., Huang, S.F., Yih, J.M., Han, T.C., 2011.
An evaluation of airlineservice quality using the fuzzy weighted
SERVQUAL method. Appl. Soft Com-put. 11 (2), 21172128.
De Oa, J., de Oa, R., Eboli, L., Mazzulla, G., 2013. Perceived
service quality in bustransit service: a structural equation
approach. Transp. Policy 29, 219226.
Dell'Olio, L., Ibeas, A., Cecin, P., 2011. The quality of
service desired by publictransport users. Transp. Policy 18 (1),
217227.
Eboli, L., Mazzulla, G., 2008. A stated preference experiment
for measuring servicequality in public transport. Transp. Plan.
Technol. 31 (5), 509523.
Eboli, L., Mazzulla, G., 2011. A methodology for evaluating
transit service qualitybased on subjective and objective measures
from the passenger's point of view.Transp. Policy 18 (1),
172181.
Eboli, L., Mazzulla, G., 2009. A new customer satisfaction index
for evaluating
transit service quality. J. Public Transp. 12 (3), 2137.
0,00
0,20
0,40
0,60
0,80
1,00V=0
V=0,1
V=0,2
V=0,3
V=0,4
V=0,5V=0,6
V=0,7
V=0,8
V=0,9
V=1
M1
M2
T1
T4
F1
Fig. 2. Sensitivity analyses ofQi values for each rail transit
lines.
0
1
2
3
4
5
6
Rank
Majority of Attributes
M1M2
T1
T4
F1
Fig. 3. The rankings of rail transit lines.
E. Celik et al. / Transport Policy 36 (2014) 283293292
http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref9http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref8http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref7http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref6http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref5http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref35http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref4http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref3http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref2http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref1
-
5/19/2018 A multiattribute customer satisfaction evaluation
approach for rail transit network A real case study for Istanbul,
Turkey.pdf
11/11
European Local Transport Information Service, 2010. Norwich Rail
station inter-change. Available from:
http://www.eltis.org/study_sheet.phtml?study_id2623&lang1enS
(accessed 06.06.14.).
Federal Transit Administration, 2010. Innovative Practices for
Increased Ridership:Capital Area Transportation Authority. United
States Department of Transpor-tation
http://ftawebprod.fta.dot.gov/BPIR/BestPractices/BP-Search.aspxS
(Ac-cessed 06.06.14.).
Filipovi, S., Tica, S., ivanovi, P., Milovanovi, B., 2009.
Comparative analysis ofthe basic features of the expected and
perceived quality of mass passengerpublic transport service in
Belgrade. Transport 24 (4), 265273.
Foote, P.J., 2004. Making buses better in Chicago: strategic
implementation ofcustomer-derived performance measures from 1995 to
2001. Transp. Res. Rec. J.Transp. Res. Board 1884 (1), 1826.
Gerek, H., Karpak, B., Klnaslan, T., 2004. A multiple criteria
approach for theevaluation of the rail transit networks in
Istanbul. Transportation 31 (2),203228.
Hassan, M.N., Hawas, Y.E., Ahmed, K., 2013. A multi-dimensional
framework forevaluating the transit service performance. Transp.
Res. Part A: Policy Pract. 50,4761.
Istanbul Ulasim A.S. (IUAS), 2013 [online]. Available from:
http://www.istanbul-ulasim.com.tr/en (accessed 25.11.13.).
Ju, Y., Wang, A., 2012. Extension of VIKOR method for
multi-criteria group decisionmaking problem with linguistic
information. Appl. Math. Model. 37, 31123125.
Kuo, M.S., 2011. A novel interval-valued fuzzy MCDM method for
improving air-lines' service quality in Chinese cross-strait
airlines. Transp. Res. Part E: Logist.Transp. Rev. 47 (6),
11771193.
Kuo, M.S., Liang, G.S., 2012. A soft computing method of
performance evaluationwith MCDM based on interval-valued fuzzy
numbers. Appl. Soft Comput. 12 (1),
476
485.Lee, L.W., Chen, S.M., 2008. Fuzzy multiple attributes group
decision-making based
on the extension of TOPSIS method and interval type-2 fuzzy
sets. In Pro-ceedings of the Seventh International Conference on
Machine Learning andCybernetic, Taipei, 32603265.
Mendel, J.M., John, R.I., Liu, F.L., 2006. Interval type-2 fuzzy
logical systems madesimple. IEEE T. Fuzzy Syst. 14 (6), 808821.
Mouwen, A., Rietveld, P., 2013. Does competitive tendering
improve customer satisfaction with public transport? A case study
for the Netherlands. Transp. ResPart A: Policy Pract. 51, 2945.
Nathanail, E., 2008. Measuring the quality of service for
passengers on the Hellenicrailways. Transp. Res. Part A: Policy
Pract. 42 (1), 4866.
Opricovic, S., Tzeng, G.H., 2004. Compromise solution by MCDM
methods: a comparative analysis of VIKOR and TOPSIS. Eur. J.
Operat. Res. 156 (2), 445455.
Opricovic, S., 1998. Multicriteria Optimization of Civil
Engineering Systems. Facultyof Civil Engineering, Belgrade.
Parasuraman, A., Zeithaml, V.A., Berry, L.L., 1988. SERVQUAL: a
multiple-item scalefor measuring perceptions of service quality. J.
Retail. 64 (Spring), 240.
Pucher, J., Kim, M., Song, J., 2005. Public transport reforms in
Seoul: innovationmotivated by funding crisis. J. Public Transp. 8
(5), 4162.
Redman, L., Friman, M., Grling, T., Hartig, T., 2013. Quality
attributes of publictransport that attract car users: a research
review. Transp. Policy 25, 119 127
Transportation Research Board, 2003. A Summary of TCRP Report
88: A Guidebookfor Developing a Transit Performance-Measurement
System. TransportationResearch Board, National Research
Council..
Turkish Statistical Institute (TUIK), 2013 [online]. Available
from: http://www.tuikgov.tr/UstMenu.do?metod temelist (accessed
25.11.13.).
Tyrinopoulos, Y., Antoniou, C., 2008. Public transit user
satisfaction: variability andpolicy implications. Transp. Policy
15, 260272. http://dx.doi.org/10.1016/jtranpol.2008.06.002.
Tzeng, G.H., Lin, C.W., Opricovic, S., 2005. Multi-criteria
analysis of alternative-fuebuses for public transportation. Energy
Policy 33 (11), 13731383.
Vahdani, B., Hadipour, H., Sadaghiani, J.S., Amiri, M., 2010.
Extension of VIKORmethod based on interval-valued fuzzy sets. Int.
J. Adv. Manuf. Technol. 47 (912), 12311239.
Wall, G., McDonald, M., 2007. Improving bus service quality and
information in
Winchester. Transp. Policy 14 (2), 165
179.Wan, S.P., Wang, Q.Y., Dong, J.Y., 2013. The extended VIKOR
method for multi-at-tribute group decision making with triangular
intuitionistic fuzzy numbersKnowl.-Based Syst. 52, 6577.
Zak, J., 2011. The methodology of multiple criteria decision
making/aiding in publictransportation. J. Adv. Transp. 45 (1),
120.
E. Celik et al. / Transport Policy 36 (2014) 283293 293
http://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://ftawebprod.fta.dot.gov/BPIR/BestPractices/BP-Search.aspxShttp://ftawebprod.fta.dot.gov/BPIR/BestPractices/BP-Search.aspxShttp://ftawebprod.fta.dot.gov/BPIR/BestPractices/BP-Search.aspxShttp://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://www.istanbul-ulasim.com.tr/enhttp://www.istanbul-ulasim.com.tr/enhttp://www.istanbul-ulasim.com.tr/enhttp://www.istanbul-ulasim.com.tr/enhttp://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref21http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref21http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref21http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://dx.doi.org/10.1016/j.tranpol.2008.06.002http://dx.doi.org/10.1016/j.tranpol.2008.06.002http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref31http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref30http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref29http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref28http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref27http://dx.doi.org/10.1016/j.tranpol.2008.06.002http://dx.doi.org/10.1016/j.tranpol.2008.06.002http://dx.doi.org/10.1016/j.tranpol.2008.06.002http://dx.doi.org/10.1016/j.tranpol.2008.06.002http://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://www.tuik.gov.tr/UstMenu.do?metod%20=temelisthttp://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref25http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref24http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref23http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref22http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref21http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref21http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref20http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref19http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref18http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref37http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref17http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref16http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref15http://www.istanbul-ulasim.com.tr/enhttp://www.istanbul-ulasim.com.tr/enhttp://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref14http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref13http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref12http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://refhub.elsevier.com/S0967-070X(14)00194-2/sbref11http://ftawebprod.fta.dot.gov/BPIR/BestPractices/BP-Search.aspxShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enShttp://www.eltis.org/study_sheet.phtml?study_id=2623&lang1=enS