Expert Systems with Applications 38 (2011) 12160–12167

Contents lists available at ScienceDirect

Expert Systems with Applications

journal homepage: www.elsevier.com/locate/eswa

A fuzzy VIKOR method for supplier selection based on entropy measure

for objective weighting
Ali Shemshadi a, Hossein Shirazi b, Mehran Toreihi a, M.J. Tarokh a,c,⇑
a
The Postgraduate IT Engineering Group, K.N. Toosi University of Technology, P.O. Box 16315-989, Tehran, Iran
b
Malek Ashtar University of Technology, Tehran, Iran
c
Industrial Engineering Faculty, K.N. Toosi University of Technology, P.O. Box 16315-989, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Keywords: Recently, resolving the problem of evaluation and ranking the potential suppliers has become as a key
Fuzzy logic strategic factor for business firms. With the development of intelligent and automated information sys-
VIKOR tems in the information era, the need for more efficient decision making methods is growing. The VIKOR
Supplier selection method was developed to solve multiple criteria decision making (MCDM) problems with conflicting and
Entropy measure
non-commensurable criteria assuming that compromising is acceptable to resolve conflicts. On the other
GMCDM
Trapezoidal fuzzy numbers
side objective weights based on Shannon entropy concept could be used to regulate subjective weights
assigned by decision makers or even taking into account the end-users’ opinions. In this paper, we treat
supplier selection as a group multiple criteria decision making (GMCDM) problem and obtain decision
makers’ opinions in the form of linguistic terms. Then, these linguistic terms are converted to trapezoidal
fuzzy numbers. We extended the VIKOR method with a mechanism to extract and deploy objective
weights based on Shannon entropy concept. The final result is obtained through next steps based on fac-
tors R, S and Q. A numerical example is proposed to illustrate an application of the proposed method.
Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction 2001; Ho, Xu, & Dey, 2010). Among these methods we can mention
artificial intelligence and knowledge discovery techniques such as
Nowadays, the problem of supplier selection has emerged as an genetic algorithm (Che & Wang, 2008; Liao & Rittscher, 2007;
active research field where numerous research papers have been Hwang & Rau, 2008), artificial neural networks (Chen, Xuan, &
published around this area within last few years. Supplier selection Shang, 2009; Lee & Ou-Yang, 2009; Wei, Zhang, & Li, 1997; Wu,
plays a key role in supply chain management (SCM) and deals with Liu, & Xi, 2008), and data mining (Kai, Xin, & Dao-ping, 2009);
evaluation, ranking and selection of the best option from a pool of mathematical programming methods such as data envelopment
potential suppliers especially in the presence of conflicting criteria. analysis (Wu, 2009), linear programming (Amid, Ghodsypour, &
Jiang, Zhuang, and Lin (2006) evinces the considerable impact of O’Brien, 2006; Guneri, Yucel, & Ayyildiz, 2009), AHP and nonlinear
supplier selection and integration on customer satisfaction and programming (Kokangul & Susuz, 2009), rough set theory (Chang,
business performance. Hung, & Lo, 2007), and grey system theory (Huixia & Tao, 2008);
With the development of information systems, it is becoming MCDM and GMCDM methods such as AHP (Chamodrakas, Batis,
an important issue for SCM frameworks and applications to be & Martakos, 2010; Lee, 2009; Xia & Wu, 2007), ANP (Gencer &
capable of making decisions on their own (Shemshadi, Soroor, & Gürpinar, 2007; Luo, Wu, Rosenberg, & Barnes, in press; Razmi,
Tarokh, 2008; Soroor, Tarokh, & Shemshadi, 2009), and it is not Rafiei, & Hashemi, in press), TOPSIS (Boran, Genc, Kurt, & Akay,
attainable until a well devised decision making process is deployed 2009; Rhee, Verma, & Plaschka, 2009); and other methods and
by an adequately improved software architecture. _
techniques (Chou & Chang, 2008; Keskin, Ilhan, & Özkan, in press;
In the literature, supplier selection has been treated as a multi- Zhang, Zhang, Lai, & Lu, 2009).
ple criteria decision making (MCDM) and a wide range of mathe- In MCDM problems, since that the valuation of criteria leads to
matical methods have been undertaken to provide the problems diverse opinions and meanings, each attribute should be imported
with sufficient and more accurate solutions (Boer, Labro, & Morlacchi, with a specific importance weight (Chen, Tzeng, & Ding, 2003). A
question rises up here and that is ‘‘how this importance weight
⇑ Corresponding author at: Industrial Engineering Faculty, K.N. Toosi University
could be calculated’’? In literature, most of the typical MCDM
of Technology, P.O. Box 16315-989, Tehran, Iran. Tel.: +98 21 88465032. methods leave this part to decision makers, while sometimes it
E-mail addresses: [email protected] (A. Shemshadi), [email protected] would be useful to engage end-users into the decision making pro-
(H. Shirazi), [email protected] (M. Toreihi), [email protected] (M.J. Tarokh). cess. To obtain a better weighting system, we may categorize

0957-4174/$ - see front matter Ó 2011 Elsevier Ltd. All rights reserved.
doi:10.1016/j.eswa.2011.03.027
 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167 12161

weighting methods into two categories: subjective methods and

objective methods (Wang & Lee, 2009). While subjective methods Non-inferior Set
determine weights solely based on the preference or judgments of
decision makers, objective methods utilize mathematical models,
such as entropy method or multiple objective programming, auto-
matically without considering the decision makers’ preferences.
The approach with objective weighting is particularly applicable
for situations where reliable subjective weights cannot be obtained Feasible Set
(Deng, Yeh, & Willis, 2000).
On the other side, new researches entail new MCDM
approaches such as VIKOR. Recently, due to its characteristics
and capabilities, the VIKOR method has been considerably under- Fig. 1. Ideal and compromise solutions.
taken by researchers to provide decision making problems, espe-
cially in the field of supplier selection, with more accurate
solutions. This includes deploying VIKOR either solely (Chiang, 3. Fuzzy logic
2009; Chen & Wang, 2009) or along with other mathematical or
MCDM approaches such as AHP (Liu & Yan,, 2007; Wu, Chen, & Fuzzy set theory first was introduced by Zadeh (1965) to map
Chen, 2010), ANP (Liou & Chuang, in press), rough sets (Jiagang & linguistic variables to numerical variables within decision making
Wei, 2008; Zhou & Tian, 2008), and artificial neural networks (Chen processes. Then the definition of fuzzy sets were manipulated to
& Li, 2008). develop Fuzzy Multi-Criteria Decision Making (FMCDM) methodol-
In this article, we provide an introduction to the VIKOR method, ogy by Bellman and Zadeh (1970) to resolve the lack of precision in
Fuzzy Logic and the Shannon Entropy respectively at sections 2–4. assigning importance weights of criteria and the ratings of alterna-
We are going to propose the new method in Section 5 while Sec- tives against evaluation criteria. The traditional logic tools gener-
tion 6 provides it with a numerical example. Section 7 concludes ally are considered outcome of bivalent (binary) logics, while
the paper. problems that pose in the real world are by no means bivalent
(Tong & Bonissone, 1980).
Just as conventional, bivalent logic is based on classic sets, fuzzy
2. VIKOR method logic is based on fuzzy sets. A fuzzy set is a set of objects in which
there is no clear-cut or predefined boundary between the objects
Vlsekriterijumska Optimizacija I Kompromisno Resenje (i.e. VI- that are or are not members of the set. A fuzzy set is characterized
KOR) method was developed by Opricovic in 1998 for multi-crite- by a membership function, which assigns to each element a grade
ria optimization of complex systems (Opricovic, 1998; Opricovic of membership within the interval [0, 1], indicating to what degree
& Tzeng, 2002). VIKOR focuses on ranking and sorting a set of that element is a member of the set (Bevilacqua, Ciarapica, & Giac-
alternatives against various, or possibly conflicting and non-com- chetta, 2006). As a result, in fuzzy logic general linguistic terms
mensurable, decision criteria assuming that compromising is such as ‘‘bad’’, ‘‘good’’ or ‘‘fair’’ could be used to capture specifically
acceptable to resolve conflicts. Similar to some other MCDM defined numerical intervals.
methods like TOPSIS, VIKOR relies on an aggregating function that A fuzzy number is defined as a fuzzy set such that
represents closeness to the ideal, but the unlike TOPSIS, intro-
M ¼ fðxÞ; lM ðxÞ; x 2 Rg ð2Þ
duces the ranking index based on the particular measure of close-
ness to the ideal solution and this method uses linear where lx(x) is a continuous mapping from R to the closed interval
normalization to eliminate units of criterion functions (Opricovic [0, 1].
& Tzeng, 2004).
In VIKOR the multi-criteria measure for compromise ranking is 3.1. Trapezoidal fuzzy numbers (TFN)
developed from the Lp  metric used as an aggregating function in a
compromise programming method (Yu, 1973; Zeleny, 1982). The A TFN can be denoted as a tuple {(n1, n2, n3, n4)jn1, n2, n3, n4 2 R;
measure Lp,j that was introduced by by Duckstein and Opricovic n1 6 n2 6 n3 6 n4} which respectively, denote the smallest possible,
(1980) represents the distance of the alternative Aj from the best the most promising, and the largest possible values that describe a
ideal solution. Each one of the various J alternatives, represented fuzzy term. Here, we can define the membership function as
as A1, A2, . . . , AJ, is measured against the ith criteria, shown by Ci, follows:
is denoted by fij. 8
The VIKOR method was developed with the form of Lp  metric, > ðx  n1 Þ=ðn2  n1 Þ; x 2 ½n1 ; n2 
>
>
<1 x 2 ½n2 ; n3 
shown as follows:
lN ðxÞ ¼ ð3Þ
> ðn4  xÞ=ðn4  n3 Þ;
> x 2 ½n3 ; n4 
( )1=p >
:
Xn
 p 0 otherwise
Lp;j ¼ wi ðfi  fij Þ=ðfi  fi Þ ð1Þ
i¼1 A TFN is shown in Fig. 2.
1 6 p 6 1; j ¼ 1; 2; . . . ; J
4. Shannon entropy and objective weights
The VIKOR method deploys L1,j (as Sj) and L1,j (as Rj) to formulate
the ranking measure. The solution obtained by minj{Sj} is with a As we mentioned before two different weights are used in the
maximum group utility and the solution obtained by minj{Rj} is proposed method: objective weights and subjective weights. Sub-
with a minimum individual regret of the opponent. The compro- jective weights could be obtained directly from the decision mak-
mise solution FC is a feasible solution that is the closest to the ideal ers’ opinions like many other MCDM processes.
F⁄, and compromise means an agreement established by mutual Shannon and Weaver (1947) proposed the entropy concept,
concessions, as illustrated in Fig. 1. which is a measure of uncertainty in information formulated in
DF 1 ¼ f1  f1C and DF 2 ¼ f2  f2C . terms of probability theory. Since the entropy concept is well
12162 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167

µ A~ ( x ) 5. The proposed method for supplier selection

In this section we are going to propose an extended version of

VIKOR that supports subjective and objective weights. The problem
of supplier selection in supply chain system could be treated as a
group multiple criteria decision making (GMCDM) problem, which
could be described by means of following sets:

1. A set of m possible alternatives called A = {A1, A2, . . ., Am};

2. A set of n decision criteria called C = {C1, C2, . . . , Cn};
3. A set of K decision makers called D = {D1, D2, . . . , DK};
x 4. A set of utility ratings of Ai (i = 1, 2, . . . , m) with respect to criteria
Cj (j = 1, 2 ,. . . , n) called X = {xij, i = 1, 2, . . . , m; j = 1, 2, . . . , n}.
Fig. 2. Trapezoidal fuzzy number.
The main steps of the proposed supplier selection algorithm
could be described as followings:
suited for measuring the relative contrast intensities of criteria to
represent the average intrinsic information transmitted to the Step 1: Identify and define linguistic terms and relevant mem-
decision maker (Zeleny, 1996), conveniently it would be a proper bership functions.
option for our purpose. Two set of appropriate linguistic variables are needed to
Later research has applied this measure to a wide range of estimate the importance weight of each criterion (Fig. 3)
applications including and the fuzzy rates of alternatives assigned by decision
makers (Fig. 4).
 Spectral analysis (Burg, 1967); Step 2: Construct a decision matrix.
 Language modeling (Rosenfeld, 1994) and Let the fuzzy rating for ith alternative regarding jth crite-
 Economics (Golan, Judge, & Miller, 1996). rion of kth decision maker be shown as ~ xijk ¼ ðxijk1 ;
xijk2 ; xijk3 ; xijk4 Þ and the importance weight of the jth crite-
Shannon developed measure H that satisfied the following rion given by the kth decision maker be w ~ sjk ¼ wsjk1 ; wsjk2 ;
properties for all pi within the estimated joint probability distribu- wsjk3 ; wsjk4 Þ. Hence, the aggregated fuzzy rating ð~ xijk Þ with
tion P (Zitnick & Kanade, 2004):

1. H is a continuous positive function;

2. If all pi are equal, pi ¼ 1n, then H should be a monotonic increas-
ing function of n; and,
3. For all, n P 2; Hðp1 ; p2 ; . . . ; pn Þ ¼ hðp1 þ p2 ; p3 ; . . . ; pn Þ þ ðp1 þ
p2 ÞH p pþp
1
; p pþp
2
1 2 1 2

He showed that the only function that satisfied these properties

is
X
HShannon ¼  pi logðpi Þ ð4Þ
i

Shannon’s concept is capable of being deployed as a weighting cal-

culation method (Lihong, Yanping, & Zhiwei, 2008; Wang & Lee,
2009), through the following steps:
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
Step 1: Normalize the evaluation index as:
Fig. 3. Linguistic variables for importance weight of each criterion.
xij
Pij ¼ P ð4-1Þ
xij
j

Step 2: Calculate entropy measure of every index using the fol-

lowing equation:

P
n
ej ¼ k Pij lnðPij Þ
j¼1 ð4-2Þ
where k ¼ ðlnðmÞÞ1
Step 3: Define the divergence through:
div j ¼ 1  ej ð4-3Þ
the more the divj is, the more important the criterion jth is!
Step 4: Obtain the normalized weights of indexes as:
div j 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 1.1
wj ¼ P ð4-4Þ
div j
j Fig. 4. Linguistic variables for the fuzzy rates of alternatives.
 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167 12163

respect to criterion Cj could be calculated as: x ~ijk ¼ measure, first we should normalize the decision matrix
fðxijk1 ; xijk2 ; xijk3 ; xijk4 Þji ¼ 1; 2; . . . ; m; j ¼ 1; 2; ::; n; k ¼ 1; for each criterion Cj (j = 1, 2, . . . , n) and calculate the pro-
2; . . . ; Kg and jection value of each criterion called Pij.
8
> xij1 ¼ minfxijk1 g xij
>
> k Pij ¼ Pm ð5-5Þ
>
> i¼1 xij
>
> P K
>
> 1
< xij2 ¼ K
> xijk2 Afterward, the entropy value can be calculated as
k¼1
ð5-1Þ follows:
>
> PK
>
> xij3 ¼ K1 xijk3
>
> X
n
1 X n
>
> k¼1
ej ¼ k Pij lnðPij Þ ¼  P ij lnðP ij Þ ð5-6Þ
>
>
: xij4 ¼ maxfxijk4 g j¼1
lnðmÞ j¼1
k

Similarly, the aggregated subjective importance Then, to calculate the degree of divergence divj of the intrinsic infor-
weightsðw ~ sj Þ could be calculated using the following rela- mation of each criterion Cj (j = 1, 2, . . . , n) we may deploy the follow-
~ sj ¼ fðwsj1 ; wsj2 ; wsj3 ; wsj4 Þjj ¼ 1; 2; ::; ng and
tions: w ing equation
8 s
>
> wj1 ¼ minfwsjk1 g div j ¼ 1  ej ð5-7Þ
>
> k
>
>
>
> ws ¼ 1 P ws
K
>
>
< j2 K jk2 The value of divj represents the inherent contrast intensity of each
k¼1
ð5-2Þ criterion Cj. The higher the divj is, the more important criterion Cj
>
> P
K
>
> ws ¼ 1 ws becomes for the problem. The objective weight for each criterion
>
> j3 K k¼1 jk3
>
> Cj can be calculated as
>
>
: wsj4 ¼ maxfwsjk4 g
k div j
woj ¼ Pn ð5-8Þ
Thus, the decision matrix De and the subjective weighting j¼1 div j
matrix Wf s could be concisely expressed as:
Step 5: Obtain the decision matrix to identify the jth criterion
2 3
x11 x12    x1n with respect to ith alternative and normalize it in order
6x    x2n 7 each criterion value is limited between 0 and 1, so the
6 21 x22 7
e ¼6
D 6 .. .. .. .. 7
7 criterions will be comparable.
4 . . . . 5 ð5-3Þ
U ¼ ½uij mn
xm1 xm2    xmn
f s ¼ ½w xþij4 ¼ maxfxij4 g; C j 2 B ð5 - 9Þ
W ~s w
~s 
1 2
~ sn 
w i

Step 3: Defuzzify the decision matrix and fuzzy weights of each xij1 ¼ minfxij1 g; C j 2 C ð5 - 10Þ
i
criterion and derive their crisp values. ( !
To derive the crisp values of arrays of decision matrix and xij1 xij2 xij3 xij4
uij ¼ ; ; ; ; Cj 2 B ð5 - 11Þ
fuzzy weights we may use the following equations: xþij4 xþij4 xþij4 xþij4
R ( !
lðxÞ:xdx xij1 xij2 xij3 xij4
defuzzðxij Þ ¼ R uij ¼ ; ; ; ; Cj 2 C ð5-12Þ
lðxÞdx xij1 xij1 xij1 xij1
R xij2  xxij1  Rx Rx x x
 
xij1 xij2 xij1
 xdx þ xij2ij3 xdx þ xij3ij4 x ij4x  xdx Step 6: Calculate the overall performance evaluation.
ij4 ij3
¼ R xij2   R R  
xxij1 x x x x
xij1 x x
dx þ xij2ij3 dx þ xij3ij4 x ij4x dx
ij2 ij1 ij4 ij3
F ¼ ½fij mn
xij1 xij2 þ xij3 xij4 þ 13 ðxij4  xij3 Þ2  13 ðxij2  xij1 Þ2  
¼ fij ¼ defuzz uij  wsj ð5-13Þ
xij1  xij2 þ xij3 þ xij4
ð5-4Þ

   
R uij2 wsj2 xwsj1 R uij3 ws R uij4 ws uij4 wsj4 x
R uij1 wsj1  xdx þ uij2 wsj3 xdx þ uij3 wsj4 uij4 ws u
uij2 wsj2 uij1 wsj1 s  xdx
lðxÞ:xdx j2 j3 j4 ij3 wj3
fij ¼ R ¼    
lðxÞdx R uij2 wsj2 xuij1 wsj1 R uij3 ws R uij4 ws uij4 wsj4 x
uij1 ws u ws u ws
dx þ uij2 wsj3 dx þ uij3 wsj4 u ws u ws
dx
j1 ij2 j2 ij1 j1 j2 j3 ij4 j4 ij3 j3

     2  2
ðuij1 uij2 Þ wsj1 wsj2 þ ðuij3 uij4 Þ wsj3 wsj4 þ 13 uij4 wsj4  uij3 wsj3  13 uij2 wsj2  uij1 wsj1
¼ ð5-14Þ
uij1 wsj1  uij2 wsj2 þ uij3 wsj3 þ uij4 wsj4

Step 4: Deploy the entropy concept to derive objective weights. Step 7: Determine the best fj and the worst fj values of all cri-
In order to determine the objective weights by entropy terion ratings.
12164 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167

fi ¼ maxffij g ð5 - 15Þ  C1: Products quality

i
 C2: Effort to establish cooperation
fi ¼ minffij g ð5-16Þ  C3: Supplier’s technical level
i
 C4: Supplier’s delay on delivery
Step 8: Compute the values of Si and Ri as follows:
 C5: Price/Cost

Xn
woj fi  fij
Si ¼   ð5 - 17Þ The committee goes through a preliminary screening and finally
j¼1
fi  fi 4(n = 4) candidate suppliers remain for further analysis.
 !
woj fi  fij
Ri ¼ max   ð5-18Þ 6.2. The solution
i fi  fi

Step 9: Compute the values Qi as follows: Step 1: The committee decides to use linguistic variables shown in
Fig. 2 to assess the importance weight of each criterion and
S ¼ maxfSi g ð5 - 19Þ the linguistic variables shown in Fig. 3 to assess the alter-
i

natives with respect to evaluation criteria. The following
S ¼ minfSi g ð5 - 20Þ tables, Tables 1 and 2, show the linguistic terms and their
i

R ¼ maxfRi g ð5 - 21Þ corresponding fuzzy numbers.

i Step 2: The decision makers deploy linguistic terms to determine

R ¼ minfRi g ð5 - 22Þ importance of each criterion and then analyze and evalu-
i
  ate each alternative with respect to evaluation criteria.
v ðSi  S Þ ð1  v ÞðRi  R Þ
Results are shown respectively in Tables 3 and 4. Linguistic
Qi ¼ þ ð5-23Þ
S  S R  R values in the two tables are converted to fuzzy numbers
using Tables 1 and 2. The results are shown in Tables 5
v is introduced as a weight for the strategy of maximum group util- and 6. Then, the aggregated fuzzy numbers according to
ity, whereas 1  v is the weight of the individual regret. the Eqs. (5-2) and (5-3) are calculated and results are
Step 10: Rank the alternatives sorting by values S, R and Q in an shown in Table 7.
ascending order
Step 11: Propose as a compromise solution the alternative (A(1)) Step 3: The aggregated fuzzy values of supplier rates are
which is the best ranked by the measure Q (minimum) defuzzyfied using (5-4) which results are shown
if the following two conditions are satisfied: in Table 8.
1. C1. Acceptable advantage: Q(A(2))  Q(A(1)) P DQ Step 4: Projection value of each criterion is calculated
Whereas A(2) with 2nd position in the ranking list by Q and using (5-5) which results are shown in Table 9.
DQ = 1/j  1. Then ej, divj and are calculated based on woj (5-
2. C2. Acceptable stability in decision making. 6), (5-7) and (5-8) relations which results are
The alternative A(1) must also be the best ranked by S or/and respectively shown in Table 10.
R. This compromise solution is stable within a decision mak- Step 5: The U matrix is determined according to (5-9),
ing process, which could be the strategy of maximum group (5-10), (5-11) and (5-12) and the result is dis-
utility (when v > 0.5 is needed), or ‘‘by consensus’’ v 0.5, played in Table 11.
or ‘‘with veto’’ v< 0.5. Here, v is the weight of decision making Step 6: The F matrix is determined in this step according
strategy of maximum group utility. to (5-13) and (5-14). Table 12 shows the result of
If one of the conditions is not satisfied, then a set of compromise (5-14) and Table 13 displays arrays of the deci-
solutions is proposed, which consists of. sion matrix F.
 Alternatives A(1) and A(m) if only the condition C2 is not
satisfied, or
 Alternatives A(1); A(2); . . .; A(M) if the condition C1 is not satis- Table 1
fied; A(M) is determined by the relation Q(A(M))  Q(A(1)) P DQ Linguistic terms and corresponding fuzzy numbers for each criterion.
for maximum M (the positions of these alternatives are ‘‘in Importance Abbreviation Fuzzy number
closeness’’).
Very low VL (0, 0, 0.1, 0.2)
Low L (0.1, 0.2, 0.2, 0.3)
Medium low ML (0.2, 0.3, 0.4, 0.5)
6. A numerical example
Medium M (0.4, 0.5, 0.5, 0.6)
Medium high MH (0.5, 0.6, 0.7, 0.8)
In this section we are going to propose a numerical example to High H (0.7, 0.8, 0.8, 0.9)
illustrate an application of the proposed method in the previous Very high VH (0.8, 0.9, 1, 1)
section. Recently, a middle-sized petrochemical factory has applied
the proposed method as its technique to evaluate, rank and select
its suppliers. We define the problem at Section 6.1 and trace the
Table 2
steps of the proposed method to provide a solution for this prob-
Linguistic terms and corresponding fuzzy numbers for each supplier.
lem at Section 6.2.
Rate Abbreviation Fuzzy number
Very poor VP (0, 0, 0.1, 0.2)
6.1. Problem definition Poor P (0.1, 0.2, 0.2, 0.3)
Medium poor MP (0.2, 0.3, 0.4, 0.5)
The company established a committee of 3(k = 3) decision mak- Fair F (0.4, 0.5, 0.5, 0.6)
ers that is responsible for supplier selection. The committee starts Medium good MG (0.5, 0.6, 0.7, 0.8)
Good G (0.7, 0.8, 0.8, 0.9)
its work with anticipation and definition of the evaluation criteria.
Very good VG (0.8, 0.9, 1, 1)
The following criteria have been defined (n = 5):
 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167 12165

Table 3 Table 8
Importance weight of criteria assessed by decision makers (linguistic values). Defuzzyfied aggregated fuzzy values of supplier rates.

C1 C2 C3 C4 C5 C1 C2 C3 C4 C5
D1 ML M H VH H A1 0.70 0.70 0.72 0.72 0.79
D2 ML MH VH VH MH A2 0.65 0.70 0.72 0.72 0.70
D3 M M VH VH MH A3 0.85 0.70 0.58 0.65 0.87
A4 0.85 0.70 0.87 0.70 0.72

Table 4 Table 9
Rating of suppliers with respect to criteria assessed by decision makers. Defuzzyfied aggregated fuzzy values of the initial supplier ratings.

C1 C2 C3 C4 C5 C1 C2 C3 C4 C5
D1 A1 MG F MG G VG A1 0.23 0.25 0.25 0.26 0.26
A2 MG MG G G MG A2 0.21 0.25 0.25 0.26 0.23
A3 G MG F MG VG A3 0.28 0.25 0.20 0.23 0.28
A4 VG G G MG G A4 0.28 0.25 0.30 0.25 0.23
D2 A1 G MG G G VG
A2 MG MG MG MG MG
A3 VG G F MG VG
A4 G MG VG G G Table 10
Calculated entropy measure, divergence and objective weights of criteria.
D3 A1 MG VG G MG MG
A2 MG G G G G C1 C2 C3 C4 C5
A3 G MG MG MG G
ej 1.00 1.00 0.99 1.00 1.00
A4 G MG VG MG MG
divj 0.00 0.00 0.01 0.00 0.00
wjo 0.32 0.00 0.46 0.04 0.17

Step 6: Table 14 displays the result of (5-15) and (5-16).

Steps 7, 8 and 9: Table 15 displays the result of (5-17), (5-18) and Step 11: Satisfied by the two conditions, A4 has the best
(5-23). rank and is chosen as the best supplier.
Step 10: Table 16.

Table 5
Importance weight of criteria assessed by decision makers (fuzzy values).

C1 C2 C3 C4 C5
D1 (0.2, 0.3, 0.4, 0.5) (0.4, 0.5, 0.5, 0.6) (0.7, 0.8, 0.8, 0.9) (0.8, 0.9, 1.0, 1.0) (0.7, 0.8, 0.8, 0.9)
D2 (0.2, 0.3, 0.4, 0.5) (0.2, 0.3, 0.4, 0.5) (0.8, 0.9, 1.0, 1.0) (0.8, 0.9, 1.0, 1.0) (0.5, 0.6, 0.7, 0.8)
D3 (0.4, 0.5, 0.5, 0.6) (0.4, 0.5, 0.5, 0.6) (0.8, 0.9, 1.0, 1.0) (0.8, 0.9, 1.0, 1.0) (0.5, 0.6, 0.7, 0.8)

Table 6
Rating of suppliers with respect to criteria assessed by decision makers (fuzzy values).

C1 C2 C3 C4 C5
D1 A1 (0.5, 0.6, 0.7, 0.8) (0.4, 0.5, 0.5, 0.6) (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9) (0.8, 0.9, 1.0, 1.0)
A2 (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8)
A3 (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.4, 0.5, 0.5, 0.6) (0.5, 0.6, 0.7, 0.8) (0.8, 0.9, 1.0, 1.0)
A4 (0.8, 0.9, 1.0, 1.0) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9)
D2 A1 (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9) (0.8, 0.9, 1.0, 1.0)
A2 (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8)
A3 (0.8, 0.9, 1.0, 1.0) (0.7, 0.8, 0.8, 0.9) (0.4, 0.5, 0.5, 0.6) (0.5, 0.6, 0.7, 0.8) (0.8, 0.9, 1.0, 1.0)
A4 (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.8, 0.9, 1.0, 1.0) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9)
D3 A1 (0.5, 0.6, 0.7, 0.8) (0.8, 0.9, 1.0, 1.0) (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8)
A2 (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9) (0.7, 0.8, 0.8, 0.9)
A3 (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8) (0.7, 0.8, 0.8, 0.9)
A4 (0.7, 0.8, 0.8, 0.9) (0.5, 0.6, 0.7, 0.8) (0.8, 0.9, 1.0, 1.0) (0.5, 0.6, 0.7, 0.8) (0.5, 0.6, 0.7, 0.8)

Table 7
Aggregated fuzzy values of supplier rates and subjective importance weights.

C1 C2 C3 C4 C5
wjs (0.20, 0.37, 0.43, 0.60) (0.20, 0.43, 0.47, 0.60) (0.70, 0.87, 0.93, 1.00) (0.80, 0.90, 1.00, 1.00) (0.50, 0.67, 0.73, 0.90)
A1 (0.50, 0.67, 0.73, 0.90) (0.40, 0.67, 0.73, 1.00) (0.50, 0.73, 0.77, 0.90) (0.50, 0.73, 0.77, 0.90) (0.50, 0.80, 0.90, 1.00)
A2 (0.50, 0.60, 0.70, 0.80) (0.50, 0.67, 0.73, 0.90) (0.50, 0.73, 0.77, 0.90) (0.50, 0.73, 0.77, 0.90) (0.50, 0.67, 0.73, 0.90)
A3 (0.70, 0.83, 0.87, 1.00) (0.50, 0.67, 0.73, 0.90) (0.40, 0.53, 0.57, 0.80) (0.50, 0.60, 0.70, 0.80) (0.70, 0.87, 0.93, 1.00)
A4 (0.70, 0.83, 0.87, 1.00) (0.50, 0.67, 0.73, 0.90) (0.70, 0.87, 0.93, 1.00) (0.50, 0.67, 0.73, 0.90) (0.50, 0.73, 0.77, 0.90)
12166 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167

Table 11
Normalized decision matrix.

uij C1 C2 C3 C4 C5
A1 (0.50, 0.67, 0.73, 0.90) (1.00, 1.67, 1.83, 2.50) (0.50, 0.73, 0.77, 0.90) (1.00, 1.47, 1.53, 1.80) (1.00, 1.60, 1.80, 2.00)
A2 (0.50, 0.60, 0.70, 0.80) (1.25, 1.67, 1.83, 2.25) (0.50, 0.73, 0.77, 0.90) (1.00, 1.47, 1.53, 1.80) (1.00, 1.33, 1.47, 1.80)
A3 (0.70, 0.83, 0.87, 1.00) (1.25, 1.67, 1.83, 2.25) (0.40, 0.53, 0.57, 0.80) (1.00, 1.20, 1.40, 1.60) (1.40, 1.73, 1.87, 2.00)
A4 (0.70, 0.83, 0.87, 1.00) (1.25, 1.67, 1.83, 2.25) (0.70, 0.87, 0.93, 1.00) (1.00, 1.33, 1.47, 1.80) (1.00, 1.47, 1.53, 1.80)

Table 12
uij  wsj matrix.

fij C1 C2 C3 C4 C5
A1 (0.06, 0.08, 0.09, 0.11) (0.13, 0.21, 0.23, 0.32) (0.13, 0.19, 0.20, 0.24) (0.28, 0.41, 0.43, 0.50) (0.21, 0.34, 0.38, 0.42)
A2 (0.06, 0.07, 0.08, 0.10) (0.16, 0.21, 0.23, 0.28) (0.13, 0.19, 0.20, 0.24) (0.28, 0.41, 0.43, 0.50) (0.21, 0.28, 0.31, 0.38)
A3 (0.08, 0.10, 0.10, 0.12) (0.16, 0.21, 0.23, 0.28) (0.11, 0.14, 0.15, 0.21) (0.28, 0.33, 0.39, 0.45) (0.30, 0.37, 0.39, 0.42)
A4 (0.08, 0.10, 0.10, 0.12) (0.16, 0.21, 0.23, 0.28) (0.18, 0.23, 0.25, 0.26) (0.28, 0.37, 0.41, 0.50) (0.21, 0.31, 0.32, 0.38)

Table 13 their work on this area and deployed a wide range of scientific
Decision matrix F. and technical techniques to enhance efficiency and flexibility of
C1 C2 C3 C4 C5
the supply networks. Supplier selection is considered as a key
problem that is generally considered as a MCDM or GMCDM prob-
A1 0.08 0.22 0.19 0.40 0.33
lem and recently has attracted many researchers to it.
A2 0.08 0.22 0.19 0.40 0.30
A3 0.10 0.22 0.15 0.36 0.37 VIKOR is a helpful tool in multi-criteria decision making, the ob-
A4 0.10 0.22 0.23 0.39 0.30 tained compromise solution could be accepted by the decision
makers because it provides a maximum group utility (represented
by min S,) of the majority, and a minimum of the individual regret
(represented by min R) of the opponent. In this article we proposed
Table 14 an extension of VIKOR that supports subjective and objective
Calculated f⁄ and f for each criterion.
weights, where subjective weights are obtained directly from the
C1 C2 C3 C4 C5 decision makers while objective weights are determined based
f⁄ 0.10 0.22 0.23 0.40 0.37 on Shannon entropy. Including end-users’ opinions will be useful
f 0.08 0.22 0.15 0.36 0.30 especially in the case of mass-customization or customer-oriented
business firms. Finally we proposed a numerical example to illus-
trate an application of the proposed method.
Table 15
The values of S, R and Q for each supplier.
References
Si A1 0.57
A2 0.74 Amid, A., Ghodsypour, S. H., & O’Brien, C. (2006). Fuzzy multiobjective linear model
A3 0.50 for supplier selection in a supply chain. International Journal of Production
A4 0.17 Economics, 104, 394–407.
Bellman, R. E., & Zadeh, L. A. (1970). Decision-making in a fuzzy environment
Ri A1 0.25 management. Science, 17(4), 141–164.
A2 0.32 Bevilacqua, M., Ciarapica, F., & Giacchetta, G. (2006). A fuzzy-QFD approach to
A3 0.46 supplier selection. Journal of Purchasing and Supply Management, 14–27.
A4 0.16 Boer, L. d., Labro, E., & Morlacchi, P. (2001). A review of methods supporting supplier
selection. European Journal of Purchasing & Supply Management, 7, 75–89.
Qi A1 0.49 Boran, F. E., Genc, S., Kurt, M., & Akay, D. (2009). A multi-criteria intuitionistic fuzzy
A2 0.77 group decision making for supplier selection with TOPSIS method. Expert
A3 0.79 Systems with Applications, 36, 11363–11368.
A4 0.00 Burg, J. (1967). Maximum entropy spectral analysis. In 37th meeting of the Society of
Exploration Geophysicists, Oklahoma City.
Chamodrakas, I., Batis, D., & Martakos, D. (2010). Supplier selection in electronic
marketplaces using satisficing and fuzzy AHP. Expert Systems with Applications,
Table 16 37, 490–498.
The ranking of suppliers in ascending order by S, R and Q. Chang, B., Hung, H. F., & Lo, C. C. (2007). Supplier selection using rough set theory. In
Proceedings of the IEEE IEEM (pp. 1461–1465).
1 2 3 4 Chen, Z. H., & Wang, H. S. (2008). Supplier selection and supply quantity allocation
of common and non-common parts with multiple criteria under multiple
By S A4 A1 A3 A2
products. Computers & Industrial Engineering, 55(1), 110–133.
By R A4 A1 A2 A3
Chen, K., & Li, J. (2008). Research on Fuzzy MCDM method based on wavelet neural
By Q A4 A1 A3 A2 network model. In International symposium on information science and
engineering (pp. 371–374).
Chen, M. F., Tzeng, G. H., & Ding, C. G. (2003). Fuzzy MCDM approach to select
service provider. In IEEE international conference on fuzzy systems (pp. 572–577).
7. Conclusion remarks Chen, K., Xuan, Z., & Shang, X. (2009). Selection of suppliers based on BP neutral
networks and grey correlation analysis. In International joint conference on
artificial intelligence (pp. 268–271).
With the today’s highly competitive business environment, Chen, L. Y., & Wang, T. C. (2009). Optimizing partners’ choice in IS/IT outsourcing
firms have to work hard and harder to reinforce their competitive projects: The strategic decision of fuzzy VIKOR. International Journal of
advantages against their rivals in market. Supply chain manage- Production Economics, 120, 233–242.
Chiang, Z. (2009). Developing an online financial decision support module based on
ment is one of the most important fields that affect manufacturers’ fuzzy mcdm method and open source tools. In International conference on
performance. Many researchers and practitioners have focused information and financial engineering (pp. 22–26).
 A. Shemshadi et al. / Expert Systems with Applications 38 (2011) 12160–12167 12167

Chou, S. Y., & Chang, Y. H. (2008). A decision support system for supplier selection Luo, X., Wu, C., Rosenberg, D., & Barnes, D. (in press). Supplier selection in agile
based on a strategy-aligned fuzzy SMART approach. Expert Systems with supply chains: An information-processing model and an illustration. Journal of
Applications, 34, 2241–2253. Purchasing & Supply Management. doi:10.1016/j.pursup.2009.05.004.
Deng, H., Yeh, C. H., & Willis, R. J. (2000). Inter-company comparison using modified Opricovic, S. (1998). Multi-criteria optimization of civil engineering systems.
TOPSIS with objective weights. Computers and Operations Research, 27, 963–973. Belgrade: Faculty of Civil Engineering.
Duckstein, L., & Opricovic, S. (1980). Multiobjective optimization in river basin Opricovic, S., & Tzeng, G.-H. (2002). Multicriteria planning of post-earthquake
development. Water Resources Research, 16(1), 14–20. sustainable reconstruction. Computer-Aided Civil and Infrastructure Engineering,,
Gencer, C., & Gürpinar, D. (2007). Analytic network process in supplier selection: A 17(3), 211–220.
case study in an electronic firm. Applied Mathematical Modelling, 31(2007), Opricovic, S., & Tzeng, G. H. (2004). Compromise solution by MCDM methods: A
2475–2486. comparative analysis of VIKOR and TOPSIS. European Journal of Operational
Golan, A., Judge, G., & Miller, D. (1996). Maximum entropy econometrics: Robust Research, 156, 445–455.
estimation with limited data. New York: John, Wiley and Sons. Razmi, J., Rafiei, H., & Hashemi, M. (in press). Designing a decision support system to
Guneri, A. F., Yucel, A., & Ayyildiz, G. (2009). An integrated fuzzy-lp approach for a evaluate and select suppliers using fuzzy analytic network process. Computers
supplier selection problem in supply chain management. Expert Systems with & Industrial Engineering. doi:10.1016/j.cie.2009.06.008.
Applications, 36, 9223–9228. Rhee, B. V. D., Verma, R., & Plaschka, G. (2009). Understanding trade-offs in the
Ho, W., Xu, X., & Dey, P. K. (2010). Multi-criteria decision making approaches for supplier selection process: The role of flexibility, delivery, and value-added
supplier evaluation and selection: A literature review. European Journal of services/support. International Journal of Production Economics, 120, 30–41.
Operational Research, 202 16–24. Rosenfeld, R. (1994). Adaptive statistical language modeling: A maximum entropy
Huixia, Z., & Tao, Y. (2008). Supplier selection model based on the Grey System approach. Ph.D. Thesis, School of Computer Science, Carnegie Mellon University.
Theory. In The International conference on risk management & engineering Shannon, C. E., & Weaver, W. (1947). The mathematical theory of communication.
management (pp. 100–104). Urbana: The University of Illinois Press.
Hwang, M. H., & Rau, H. (2008). Development of a supplier selection approach from Shemshadi, A., Soroor, J., & Tarokh, M. J. (2008). Implementing a multi-agent system
the viewpoint of the entire supply chain. In Proceedings of the seventh for the real-time coordination of a typical supply chain based on the JADE
international conference on machine learning and cybernetics (pp. 3938–3945). Technology. In 3rd IEEE SMC international conference on system of systems
Jiagang, G., & Wei, Z. (2008). Selection of suppliers based on rough set theory and engineering (pp. 1–6).
VIKOR algorithm. In International symposium on intelligent information Soroor, J., Tarokh, M. J., & Shemshadi, A. (2009). Initiating a state of art system for
technology application workshops, doi:10.1109/IITA.Workshops.2008.213. real-time supply chain coordination. European Journal of Operational Research,
Jiang, Z. F., Zhuang, T. Y., & Lin, S. X. (2006). Empirical analysis of the effects of 196(2), 635–650.
supplier selection and integration on customer satisfaction and business Tong, R., & Bonissone, P. (1980). A linguistic approach to decisionmaking with fuzzy
performance. In IEEE International Conference on management of innovation sets. IEEE Transactions On Systems Man Cybernetics SMC, 10(11), 716–723.
and technology (pp. 931–935). Wang, T. C., & Lee, H. D. (2009). Developing a fuzzy TOPSIS approach based on
Kai, Z., Xin, Y., & Dao-ping, W. (2009). Study on CBR supplier selection system based subjective weights and objective weights. Expert Systems with Applications, 36,
on data mining for oil enterprises. In International symposium on information 8980–8985.
engineering and electronic commerce (pp. 555–559). Wei, S., Zhang, J., & Li, Z. (1997). A supplier-selecting system using a neural network.
_
Keskin, G.A., Ilhan, S., & Özkan, C. (in press). The fuzzy ART algorithm: A In IEEE international conference on intelligent processing systems (pp. 468–471).
categorization method for supplier evaluation and selection. Expert Systems Wu, D. (2009). Supplier selection: A hybrid model using DEA, decision tree and
with Applications. doi:10.1016/j.eswa.2009.06.004. neural network. Expert Systems with Applications, 36, 9105–9112.
Kokangul, A., & Susuz, Z. (2009). Integrated analytical hierarch process and Wu, H. Y., Chen, J. K., & Chen, I. S. (2010). Innovation capital indicator assessment of
mathematical programming to supplier selection problem with quantity Taiwanese Universities: A hybrid fuzzy model application. Expert Systems with
discount. Applied Mathematical Modelling, 33, 1417–1429. Applications, 37, 1635–1642.
Lee, A. H. I. (2009). A fuzzy supplier selection model with the consideration of Wu, J. J., Liu, G., & Xi, C. Y. (2008). The study on agile supply chain-based supplier
benefits, opportunities, costs and risks. Expert Systems with Applications, 36, selection and evaluation. In Int. Symposium on Information Science and
2879–2893. Engineering, 280-284.
Lee, C. C., & Ou-Yang, C. (2009). A neural networks approach for forecasting the Xia, W., & Wu, Z. (2007). Supplier selection with multiple criteria in volume
supplier’s bid prices in supplier selection negotiation process. Expert Systems discount environments. Omega, 35, 494–504.
with Applications, 36, 2961–2970. Yu, P. L. (1973). A class of solutions for group decision problems. Management
Liao, Z., & Rittscher, J. (2007). Integration of supplier selection, procurement lot Science, 19(8), 936–946.
sizing and carrier selection under dynamic demand conditions. International Zadeh, L. A. (1965). Fuzzy sets. Information Control, 8, 338–353.
Journal of Production Economics, 107, 502–510. Zeleny, M. (1982). Multiple criteria decision making. New York: Mc-Graw-Hill.
Lihong, M., Yanping, Z., & Zhiwei, Z. (2008). Improved VIKOR algorithm based on Zeleny, M. (1996). Multiple criteria decision making. New York: Springer.
AHP and Shannon entropy in the selection of thermal power enterprise’s coal Zhang, D., Zhang, J., Lai, K. K., & Lu, Y. (2009). An novel approach to supplier selection
suppliers. In International conferenace on information management, innovation based on vague sets group decision. Expert Systems with Applications, 36,
management and industrial engineering (pp. 129–133). 9557–9563.
Liou, J. J. H., & Chuang, Y. T. (in press). Developing a hybrid multi-criteria model for Zhou, J., & Tian, J. (2008). Optial selection of FGD technology using GA-based rough
selection of outsourcing providers. Expert Systems with Applications. sets theory and VIKOR. In Proceedings of the IEEE international conference on
doi:10.1016/j.eswa.2009.11.048. automation and logistics (pp. 1276–1280).
Liu, H., & Yan, T. (2007). Bidding-evaluation of construction projects based on VIKOR Zitnick, C. L., & Kanade T. (2004). Maximum entropy for collaborative filtering. In
method. In Proceedings of the IEEE international conference on automation and ACM proceedings of the 20th conference on uncertainty in artificial intelligence (pp.
logistics (pp. 1778–1782). 636–643).