Information Sciences 176 (2006) 2405–2416

www.elsevier.com/locate/ins

Ranking of fuzzy numbers by sign distance

a,b,* c
S. Abbasbandy , B. Asady
a
Department of Mathematics, Science and Research Branch, Islamic Azad University,
Tehran 14515, Iran
b
Department of Mathematics, Imam Khomeini International University, Ghazvin 34194, Iran
c
Department of Mathematics, Arak Branch, Islamic Azad University, Arak 38135, Iran

Received 24 October 2003; received in revised form 8 March 2005; accepted 21 March 2005

Abstract

Several different strategies have been proposed for ranking of fuzzy numbers. These
include methods based on the coefficient of variation (CV index), distance between fuzzy
sets, centroid point and original point, and weighted mean value. Each of these tech-
niques has been shown to produce non-intuitive results in certain cases. In this paper
we propose a modification of the distance based approach called the sign distance, which
is both efficient to evaluate and able to overcome the shortcomings of the previous tech-
niques. The calculation of the proposed method is far simpler than the other approaches.
Ó 2005 Elsevier Inc. All rights reserved.

Keywords: Fuzzy numbers; Ranking of fuzzy numbers; Sign distance

1. Introduction

In many applications, ranking of fuzzy numbers is an important component

of the decision process. Following the pioneering work of Jain [10,11] and

*
Corresponding author. Address: Department of Mathematics, Science and Research Branch,
Islamic Azad University, Tehran 14515, Iran. Tel.: +98 9121 305326; fax: +98 2813 664099.
E-mail address: [email protected] (S. Abbasbandy).

0020-0255/$ - see front matter Ó 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.ins.2005.03.013
2406 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

Dubois and Prade [8], who used maximizing sets to order fuzzy numbers,
numerous ranking techniques have been proposed and investigated. Some of
them have been compared and contrasted in Bortolan and Degani [3], and
more recently in Chu and Tsao [7]. Cheng [5] proposed the distance method
for ranking of fuzzy numbers, i.e.,
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
RðuÞ ¼ x2 þ y 2 ;
where
Rb Rc Rd R1 R1
xuL dx þ x dx þ xuR dx 0
ru dr þ ru dr
a
x ¼ R b b
Rc R dc ; y ¼ R 1 R01 ;
a
uL dx þ b
dx þ c
uR dx 0
u dr þ 0
u dr
uL, uR are the left and right membership functions of fuzzy number u, and ðu; uÞ
is the parametric form (see Definitions 1.1 and 1.2). The resulting scalar
value R(u) is used to rank the fuzzy numbers; if R(ui) < R(uj), then ui  uj. If
R(ui) > R(uj), then ui  uj; if R(ui) = R(uj), then ui  uj. Chu and Tsao [7] re-
viewed ChengÕs method [5] and claimed that it has shortcomings. For example,
consider three triangular fuzzy numbers, u1 = (0.3, 0.1, 0.2), u2 = (0.32, 0.15,
0.26), and u3 = (0.4, 0.15, 0.3) from [5,7]. By ChengÕs distance method, R(u1) =
0.590, R(u2) = 0.604, and R(u3) = 0.662, producing the ranking order
u1  u2  u3 (see Fig. 1). Consequently, we can logically infer that the ranking
order of the images of these fuzzy numbers (opposite with respect to origin, [7])
is u1  u2  u3. However, by distance method, the ranking order remains
u1  u2  u3. Consequently, it also has shortcomings.

u1
0.9

0.8

0.7

0.6

0.5

0.4 u3

0.3
u2
0.2 u2 u3
u1

0.1

0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8

Fig. 1. Triangular fuzzy numbers u1 = (0.3, 0.1, 0.2), u2 = (0.32, 0.15, 0.26) and u3 = (0.4, 0.15, 0.3).
 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416 2407

Cheng [5] proposed the coefficient of variance (CV index), i.e. CV = r (stan-
dard error)/jlj (mean), l 5 0, r > 0. In this approach, the fuzzy number with
smaller CV index is ranked higher. In Table 1, sets 1, 2 and 4, we illustrate that
ChengÕs CV index also contains shortcomings.
Chu and Tsao [7] proposed the area between the centroid point and
original point for ranking; i.e. SðuÞ ¼ x y . This method for some fuzzy num-
bers is unreasonable (for more details Rsee Example 2). R 1 Yager [18] proposed
1
weighted mean value (or centroid, 1 xlu ðxÞ dx= 1 lu ðxÞ dx) to define
ordering.
Having reviewed the previous methods, we now turn to introduce our pro-
posed approach, termed as sign distance method. We consider a fuzzy origin
for fuzzy numbers, then according to the distance of fuzzy numbers with re-
spect to this origin we rank them.
The basic definitions of a fuzzy number are given in [9,20,21] as follows.

Definition 1.1. A fuzzy number is a fuzzy set like u : R ! I ¼ ½0; 1 which

satisfies:

1. u is upper semi-continuous,
2. u(x) = 0 outside some interval [a, d],
3. There are real numbers a, b such that a 6 b 6 c 6 d and
(a) u(x) is monotonic increasing on [a, b],
(b) u(x) is monotonic decreasing on [c, d],
(c) u(x) = 1, b 6 x 6 c.

The membership function u can be expressed as

8
> uL ðxÞ a 6 x 6 b;
>
>
<1 b 6 x 6 c;
uðxÞ ¼
>
> uR ðxÞ c 6 x 6 d;
>
:
0 otherwise;

where uL : ½a; b ! ½0; 1 and uR : ½c; d ! ½0; 1 are left and right membership
functions of fuzzy number u. An equivalent parametric form is also given in
[12] as follows.

Definition 1.2. A fuzzy number u in parametric form is a pair ðu; uÞ of func-
uðrÞ, 0 6 r 6 1, which satisfy the following requirements:
tions uðrÞ; 

1. u(r) is a bounded monotonic increasing left continuous function,

uðrÞ is a bounded monotonic decreasing left continuous function,
2. 
3. uðrÞ 6 uðrÞ, 0 6 r 6 1.
2408 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

Table 1
Comparative results of Example 1
Authors Fuzzy Set 1 Set 2 Set 3 Set 4
number
Choobineh and Li A 0.333 0.458 0.333 0.50
B 0.50 0.583 0.4167 0.5833
C 0.667 0.667 0.5417 0.6111
Results ABC ABC ABC ABC
A 0.60 0.575 0.5 0.45
Yager B 0.70 0.65 0.55 0.525
C 0.80 0.7 0.625 0.55
Results ABC ABC ABC ABC
A 0.3375 0.4315 0.375 0.52
Chen B 0.50 0.5625 0.425 0.57
C 0.667 0.625 0.55 0.625
Results ABC ABC ABC ABC
Baldwin and Guild A 0.30 0.27 0.27 0.40
B 0.33 0.27 0.37 0.42
C 0.44 0.37 0.45 0.42
Results ABC ABC ABC ABC
Chu and Tsao A 0.299 0.2847 0.25 0.24402
B 0.350 0.32478 0.31526 0.26243
C 0.3993 0.350 0.27475 0.2619
Results ABC ABC ACB ACB
Yao and Wu A 0.6 0.575 0.5 0.475
B 0.7 0.65 0.625 0.525
C 0.8 0.7 0.55 0.525
Results ABC ABC ACB ABC
Sign distance method p = 1 A 1.2 1.15 1 0.95
B 1.4 1.3 1.25 1.05
C 1.6 1.4 1.1 1.05
Results ABC ABC ACB ABC
Sign distance method p = 2 A 0.8869 0.8756 0.7257 0.7853
B 1.0194 0.9522 0.9416 0.7958
C 1.1605 1.0033 0.8165 0.8386
Results ABC ABC ACB ABC
Cheng distance A 0.79 0.7577 0.7071 0.7106
B 0.8602 0.8149 0.8037 0.7256
C 0.9268 0.8602 0.7458 0.7241
Results ABC ABC ACB ACB
Cheng CV uniform A 0.0272 0.0328 0.0133 0.0693
distribution B 0.0214 0.0246 0.0304 0.0385
C 0.0225 0.0095 0.0275 0.0433
Results BCA CBA ACB BCA
 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416 2409

Table 1 (continued)
Authors Fuzzy Set 1 Set 2 Set 3 Set 4
number
Cheng CV proportional A 0.0183 0.026 0.008 0.0471
distribution B 0.0128 0.0146 0.0234 0.0236
C 0.0137 0.0057 0.0173 0.0255
Results BCA CBA ACB BCA

The trapezoidal fuzzy number u = (x0, y0, r, b), with two defuzzifier x0, y0,
and left fuzziness r > 0 and right fuzziness b > 0 is a fuzzy set where the
membership function is as
81
>
> ðx  x0 þ rÞ x0  r 6 x 6 x0 ;
>r
>
<1 x 2 ½x0 ; y 0 ;
uðxÞ ¼
> 1 ðy 0  x þ bÞ
> y 0 6 x 6 y 0 þ b;
>
> b
:
0 otherwise;

and its parametric form is

uðrÞ ¼ x0  r þ rr; uðrÞ ¼ y 0 þ b  br:


The addition and scalar multiplication of fuzzy numbers are defined by the
extension principle and can be equivalently represented in [9,20,21] as follows.
For arbitrary u ¼ ðu; 
uÞ, v ¼ ðv; vÞ we define addition (u + v) and multipli-
cation by real scalar k > 0 as

ðu þ vÞðrÞ ¼ uðrÞ þ vðrÞ; ðu þ vÞðrÞ ¼ 

uðrÞ þ vðrÞ; ð1Þ

ðkuÞðrÞ ¼ kuðrÞ; ðkuÞðrÞ ¼ k

uðrÞ: ð2Þ

The collection of all fuzzy numbers with addition and multiplication as defined
by (1) and (2) is denoted by E, which is a convex cone. The image (opposite) of
u = (x0, y0, r, b), can be defined by u = (y0, x0, b, r) (see [19–21]).

Definition 1.3. For arbitrary fuzzy numbers u ¼ ðu; uÞ and v ¼ ðv; vÞ, the
function
Z 1 Z 1 1=p
p p
Dp ðu; vÞ ¼ juðrÞ  vðrÞj dr þ j
uðrÞ  vðrÞj dr ðp P 1Þ
0 0

is the distance between u and v (see [1,13,14]).

2410 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

2. Ranking of fuzzy numbers by sign distance

In this section, we will propose the ranking of fuzzy numbers associated with
the metric D in E.
The membership function of a 2 R is ua(x) = 1, if x = a, and ua(x) = 0, if
x 5 a. Hence if a = 0, we will have

1 x ¼ 0;
u0 ðxÞ ¼
0 x 6¼ 0:
We consider u0 as a fuzzy origin and since u0 2 E, left fuzziness r and right
fuzziness b are 0, so for each u 2 E
Z 1 1=p
p p
Dp ðu; u0 Þ ¼ ðjuðrÞj þ j
uðrÞj Þ dr ðp P 1Þ: ð3Þ
0

Definition 2.1. Let c : E ! f1; 1g be a function that is defined as follows:

Z 1 
8u 2 E : cðuÞ ¼ sign ðuðrÞ þ 
uðrÞÞ dr ;
0

where
8 R 
>
<1 if sign
1
ðu þ 
uÞðrÞ dr P 0;
0
cðuÞ ¼ R 
>
: 1 1
if sign 0 ðu þ  uÞðrÞ dr < 0:

Remark 2.1. If inf supp(u) P 0 or inf u(r) P 0 then c(u) = 1.

uðrÞ < 0 then c(u) = 1.

Remark 2.2. If sup supp(u) < 0 or sup 

Definition 2.2. For u 2 E,

d p ðu; u0 Þ ¼ cðuÞDp ðu; u0 Þ; ð4Þ
is called sign distance.

Definition 2.3. For u and v 2 E, define the ranking of u and v by dp on E, i.e.

d p ðu; u0 Þ > d p ðv; u0 Þ if and only if u  v;

d p ðu; u0 Þ > d p ðv; u0 Þ if and only if u  v;

d p ðu; u0 Þ ¼ d p ðv; u0 Þ if and only if u  v:

 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416 2411

Then we formulate the order Ô%Õ and Ô†Õ as

u%v if and only if u  v or u  v;

u-v if and only if u  v or u  v:

We consider the following reasonable properties for the ordering
approaches, see [15,16].

A1: For an arbitrary finite subset C of E and u 2 C, u % u.

A2: For an arbitrary finite subset C of E and (u, v) 2 C2, u % v and v % u, we
should have u  v.
A3: For an arbitrary finite subset C of E and (u, v, z) 2 C3, u % v and v % z,
we should have u % z.
A4: For an arbitrary finite subset C of E and (u, v) 2 C2, inf supp(u) >
sup supp(v), we should have u % v.
A04 : For an arbitrary finite subset C of E and (u, v) 2 C2, inf supp(u) >
sup supp(v), we should have u  v.
A5: Let C and C 0 be two arbitrary finite subsets of E in which u and v are in
C \ C 0 . We obtain the ranking order u  v by dp on C 0 if and only if u  v by
dp on C.
A6: Let u, v, u + z and v + z be elements of E. If u % v, then u + z % v + z.
A06 : Let u, v, u + z and v + z be elements of E. If u  v, then u + z  v + z.

Remark 2.3. The function dp, sign distance, has the properties A1, A2, . . . , A5.

Remark 2.4. The function dp, sign distance, for p = 1 has the properties A6, A06
if inf{supp(u), supp(v), supp(u + z), supp(v + z)} P 0 or sup{supp(u), supp(v),
supp(u + z), supp(v + z)} 6 0.

Remark 2.5. Suppose u and v 2 E are arbitrary, then

(i) If u = v then u  v,
p p p p
(ii) If v  u and cðuÞðjuðrÞj þ j
uðrÞj Þ > cðuÞðjvðrÞj þ jvðrÞj Þ for all r 2 [0, 1]
then v  u.

Remark 2.6. If u  v, it is not necessary that u = v. Since if u 5 v and

cðuÞðjuðrÞjp þ juðrÞjp Þ ¼ cðuÞðjvðrÞjp þ jvðrÞjp Þ then u  v.

Remark 2.7. If u  v then u v.

Therefore we can simply rank the fuzzy numbers by the defuzzification of
dp(u, u0). By Remark 2.5 we can logically infer ranking order of the images
of the fuzzy numbers.
2412 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

Now we compare our method with the others in [2,4,6,7,17,18].

Example 1. Consider the following sets, see Yao and Wu [17].

Set 1: A = (0.5, 0.1, 0.5), B = (0.7, 0.3, 0.3), C = (0.9, 0.5, 0.1).
Set 2: A = (0.4, 0.7, 0.1, 0.2) (trapezoidal fuzzy number), B = (0.7, 0.4, 0.2),
C = (0.7, 0.2, 0.2).

0.9

0.8

0.7

0.6

0.5 B C
A B C A
0.4

0.3

0.2

0.1

0
0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 2. Set 1.

0.9

0.8

0.7

0.6
A
0.5
A B C B
0.4
C
0.3

0.2

0.1

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 3. Set 2.
 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416 2413

Set 3: A = (0.5, 0.2, 0.2), B = (0.5, 0.8, 0.2, 0.1), C = (0.5, 0.2, 0.4).
Set 4: A = (0.4, 0.7, 0.4, 0.1), B = (0.5, 0.3, 0.4), C = (0.6, 0.5, 0.2).

In Table 1 we have the following results: In set 1, the CV index, the ranking
order is B  C  A, for both CV distributions, wherein the result is unreason-
able (see Fig. 2). But the ranking order for our method and other methods is

B
1

0.9 A

0.8
C
0.7 B

0.6

0.5
B
0.4
A
0.3 C

0.2

0.1

0
0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

Fig. 4. Set 3.

0.9

0.8 C A
B
0.7

0.6

0.5

0.4 A

0.3
B
0.2 C
0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9

Fig. 5. Set 4.
2414 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

A  B  C. In set 2, Baldwin and Guild method, the ranking order is

A  B  C and C  B  A is gained for CV uniform distribution and CV pro-
portional distribution, which are all shortcomings of these methods. But our
method has the same result as other six techniques, which have not above
shortcomings (see Fig. 3). However in set 3, we have A  C  B, which is
the same as the results of ChengÕs technique and Yao–Wu and Chu–Tsao
methods, but all others have A  B  C. We conclude from Fig. 4 that
A  C  B is better than A  B  C. In set 4, Fig. 5, A  B  C for sign dis-
tance method by p = 2 leads to the same results of Choobineh–Li, Yager and
Chen. The results for Baldwin–Guild and Yao–Wu methods are A  B  C,
the same as our method for p = 1. Thus ranking order for ChengÕs technique,
in the distance method and CV index is A  C  B and B  C  A, respec-
tively and the result of Chu–Tsao method is A  C  B.

Table 2
Comparative results of Example 2
Fuzzy Sign Sign Chu and Tsao Cheng distance CV index
number distance p = 1 distance p = 2
A 6.12 8.52 3 6.021 0.028
B 12.45 8.82 3.126 6.349 0.0098
C 12.5 8.85 3.085 6.3519 0.0089
Results ABC ABC ACB ABC CBA

0.9

0.8

0.7
A
0.6
B
0.5 C

0.4 A B C

0.3

0.2

0.1

0
5 5.2 5.4 5.6 5.8 6 6.2 6.4 6.6 6.8 7

Fig. 6. Example 2.
 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416 2415

Example 2. Consider the three triangular fuzzy numbers, A = (6, 1, 1),

B = (6, 0.1, 1) and C = (6, 0, 1). The results are given in Table 2.
In Table 2 the results of Chu–Tsao method and Cheng CV index are unrea-
sonable. The results of sign distance methods are the same as Cheng distance
method, i.e., A  B  C, see Fig. 6.

3. Conclusions

In spite of many ranking methods, no one can rank fuzzy numbers with
human intuition consistently in all cases. Shortcomings are found in ranking
fuzzy numbers with the coefficient of variation (CV index), distance between
fuzzy sets, centroid point and original point, and weighted mean value. To
overcome shortcomings we proposed sign distance method. It can effectively
rank various fuzzy numbers and their images. The calculations of the proposed
method is far simpler than the other approaches. Finally, comparative exam-
ples were presented to illustrate the advantage of the sign distance method.

Acknowledgments

The authors wish to express their thanks to the referees for comments which
improved the paper.

References

[1] S. Abbasbandy, B. Asady, A note on ‘‘A new approach for defuzzification’’, Fuzzy Sets Syst.
128 (2002) 131–132.
[2] J.F. Baldwin, N.C.F. Guild, Comparison of fuzzy numbers on the same decision space, Fuzzy
Sets Syst. 2 (1979) 213–233.
[3] G. Bortolan, R. Degani, A review of some methods for ranking fuzzy numbers, Fuzzy Sets
Syst. 15 (1985) 1–19.
[4] S. Chen, Ranking fuzzy numbers with maximizing set and minimizing set, Fuzzy Sets Syst. 17
(1985) 113–129.
[5] C.H. Cheng, A new approach for ranking fuzzy numbers by distance method, Fuzzy Sets Syst.
95 (1998) 307–317.
[6] F. Choobineh, H. Li, An index for ordering fuzzy numbers, Fuzzy Sets Syst. 54 (1993) 287–
294.
[7] T. Chu, C. Tsao, Ranking fuzzy numbers with an area between the centroid point and original
point, Comput. Math. Appl. 43 (2002) 11–117.
[8] D. Dubois, H. Prade, Operations on fuzzy numbers, Int. J. Syst. Sci. 9 (1978) 626–631.
[9] D. Dubois, H. Prade, Fuzzy Sets and Systems: Theory and Application, Academic Press, New
York, 1980.
2416 S. Abbasbandy, B. Asady / Information Sciences 176 (2006) 2405–2416

[10] R. Jain, Decision-making in the presence of fuzzy variable, IEEE Trans. Syst. Man Cybernet. 6
(1976) 698–703.
[11] R. Jain, A procedure for multi-aspect decision making using fuzzy sets, Int. J. Syst. Sci. 8
(1978) 1–7.
[12] M. Ma, M. Friedman, A. Kandel, A new fuzzy arithmetic, Fuzzy Sets Syst. 108 (1999) 83–90.
[13] M. Ma, A. Kandel, M. Friedman, Correction to ‘‘A new approach for defuzzification’’, Fuzzy
Sets Syst. 128 (2002) 133–134.
[14] M. Ma, A. Kandel, M. Friedman, A new approach for defuzzification, Fuzzy Sets Syst. 111
(2000) 351–356.
[15] X. Wang, E.E. Kerre, Reasonable properties for the ordering of fuzzy quantities I, Fuzzy Sets
Syst. 118 (2001) 375–385.
[16] X. Wang, E.E. Kerre, Reasonable properties for the ordering of fuzzy quantities II, Fuzzy Sets
Syst. 118 (2001) 387–405.
[17] J. Yao, K. Wu, Ranking fuzzy numbers based on decomposition principle and signed distance,
Fuzzy Sets Syst. 116 (2000) 275–288.
[18] R.R. Yager, A procedure for ordering fuzzy subsets of the unit interval, Inform. Sci. 24 (1981)
143–161.
[19] A. Kauffman, M.M. Gupta, Introduction to Fuzzy Arithmetic: Theory and Application, Van
Nostrand Reinhold, New York, 1991.
[20] L.A. Zadeh, Fuzzy sets, Inform. Control 8 (1965) 338–353.
[21] H.J. Zimmermann, Fuzzy Sets Theory and Its Application, Kluwer Academic Press,
Dordrecht, 1991.