Applied Mathematics and Computation 262 (2015) 191–198

Contents lists available at ScienceDirect

Applied Mathematics and Computation

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

Ranking generalized exponential trapezoidal fuzzy numbers

based on variance
S. Rezvani∗
Department of Mathematics, Imam Khomaini Mritime University of Nowshahr, Nowshahr, Iran

a r t i c l e i n f o a b s t r a c t
MSC: In this paper, we calculate ranking of exponential trapezoidal fuzzy numbers based on variance.
47S40 In this method, values are calculated by finding expected values using the probability density
08A72
function corresponding to the membership functions of the given fuzzy number and provides
03E72
the correct ordering of exponential trapezoidal fuzzy numbers and also this approach is very
Keywords: simple and easy to apply in the real life problems. For the validation, the results of the approach
Exponential trapezoidal fuzzy numbers are compared with different existing approaches.
Ranking method
© 2015 Published by Elsevier Inc.
Variance

1. Introduction

In most of the cases in our life, the data obtained for decision making are only approximately known. In 1965, Zadeh [27]
introduced the concept of fuzzy set theory to meet those problems. In 1978, Dubois and Prade defined any of the fuzzy number
as a fuzzy subset of the real line [10]. Fuzzy numbers allow us to make the mathematical model of linguistic variable or fuzzy
environment. Ranking fuzzy numbers were first proposed by Jain [11] for decision making in fuzzy situations by representing
the ill-defined quantity as a fuzzy set. Bortolan and Degani [3] reviewed some of these ranking methods [2,6,10,11,14,15] for
ranking fuzzy subsets. Chen [4] presented ranking fuzzy numbers with maximizing set and minimizing set. [12] and Wang and
Lee [26] also used the centroid concept in developing their ranking index. Chen and Chen [5] presented a method for ranking
generalized trapezoidal fuzzy numbers. Abbasbandy and Hajjari [1] introduced a new approach for ranking of trapezoidal fuzzy
numbers based on the left and right spreads at some -levels of trapezoidal fuzzy numbers. Chen and Chen [7] presented a method
for fuzzy risk analysis based on ranking generalized fuzzy numbers with different heights and different spreads and S. K. Barik [3]
presented a Probabilistic Quadratic Programming Problems with some fuzzy parameters. Rezvani [14–23] evaluated the system
of ranking and properties of fuzzy numbers. Moreover, Rezvani [16] proposed a new method for ranking in perimeters of two
generalized trapezoidal fuzzy numbers. In this paper, we calculate ranking of exponential trapezoidal fuzzy numbers based on
variance. In this method, values are calculated by finding expected values using the probability density function corresponding
to the membership functions of the given fuzzy number and provides the correct ordering of exponential trapezoidal fuzzy
numbers and also this approach is very simple and easy to apply in the real life problems. For the validation, the results of the
approach are compared with different existing approaches.

∗
Tel.: +989374433398.
E-mail address: [email protected], [email protected]

http://dx.doi.org/10.1016/j.amc.2015.04.030
0096-3003/© 2015 Published by Elsevier Inc.
192 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198

2. Preliminaries

Generally, a generalized fuzzy number A is described as any fuzzy subset of the real line R, whose membership function μA
satisfies the following conditions,

(i) μA is a continuous mapping from R to the closed interval [0,1],

(ii) μA (x) = 0, − < u  a,
(iii) μA (x) = L(x) is strictly increasing on [a, b],
(iv) μA (x) = w, b  x  c,
(v) μA (x) = R(x) is strictly decreasing on [c, d],
(vi) μA (x) = 0, d  x < 
Where 0 < w  1 and a, b, c, and d are real numbers. We call this type of generalized fuzzy number a trapezoidal fuzzy number,
and it is denoted bye A = (a, b, c, d; w)LR . When w = 1, this type of generalized fuzzy number is called normal fuzzy number and
is represented by A = (a, b, c, d)LR .
However, these fuzzy numbers always have a fix range as [c, d] . Here, we define its general form as follows:
⎧ −[(b−x)/(b−a)]
⎨ we a ≤ x ≤ b,
fA (x) = w b ≤ x ≤ c, (1)
⎩ −[(x−c)/(d−c)]
we c ≤ x ≤ d,
where 0 < w  1, a, b are real numbers, and c, d are positive real numbers. we denote this type of generalized exponential fuzzy
number as A = (a, b, c, d; w)E .
Especially, when w = 1, we denote it as A = (a, b, c, d)E . we define the representation of generalized exponential fuzzy number
based on the integral value of graded mean h-level as follow. Let the generalized exponential fuzzy number A = (a, b, c, d)E , where
0 < w  1, and c, d are positive real numbers, a, b are real numbers as in formula (2.1). Now, let two monotonic functions be
L(x) = we−[(b−x)/(b−a)] , R(x) = we−[(x−c)/(d−c)] (2)

3. Proposed approach

As we know that any probability density function with finite support is associated with an expected value, the Mellin transform
[24] is used to find this expected value.

Definition 1. The Mellin transform MX (t) of a probability density function f(x), where x is a positive, is defined as
 ∞
MX (t) = xt−1 f (x) dx (3)
0

where the integral exists.

Now we find the Mellin transform in terms of expected values. Recall that the expected value of any function g(X) of the
random variable X, whose probability density function is f(x), is given by
 ∞
E[g(X )] = g(x) f (x) dx (4)
−∞
∞
Therefore, it follows that MX (t) = E[X t−1 ] = 0 xt−1 f (x) dx.
Hence, E[Xt ] = MX (t + 1). Thus, the expected value of random variable X is E[x] = MX (2).
Let A = (a, b, c, d)E generalized exponential fuzzy number, then the crisp values are calculated by finding expected values
using the probability density function corresponding to the membership functions of the given fuzzy number.

Theorem 1. The probability density function corresponding to exponential trapezoidal fuzzy number A = (a, b, c, d)E is given as
fA (x) = C μA (x) (5)
where μA (x) is defined as
⎧ −[(b−x)/(b−a)]
⎨ we a ≤ x ≤ b,
μA (x) = w b ≤ x ≤ c, (6)
⎩ −[(x−c)/(d−c)]
we c ≤ x ≤ d,
Proof. C is calculated as
 ∞
fA (x) dx = 1 (7)
−∞

that is,
 ∞
C μA (x) dx = 1 (8)
−∞
 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198 193

that is,
 b  c  d
C e−[(b−x)/(b−a)] dx + C dx + C e−[(x−c)/(d−c)] = 1 (9)
a b c

On integration, we get
1 1
C [(b − a)(1 − )] + C (c − b) + C [(c − d)( − 1)] = 1
e e
e
⇒C= (10)
a(1 − e) − b + c + (e − 1)d
The proportional probability density function corresponding to exponential trapezoidal fuzzy number A is given by
⎧
⎪ we we[(x−a)/(b−a)]
⎪
⎪
⎪ e−[(b−x)/(b−a)] = a ≤ x ≤ b,
⎪
⎪ a(1 − e) − b + c + (e − 1)d a(1 − e) − b + c + (e − 1)d
⎨ we
fA (x) = b ≤ x ≤ c, (11)
⎪
⎪ a(1 − e) − b + c + (e − 1)d
⎪
⎪ ( ) ( )
⎪
⎪ we we [ d−x / d−c ]
⎩ e−[(x−c)/(d−c)] = c ≤ x ≤ d,
a(1 − e) − b + c + (e − 1)d a(1 − e) − b + c + (e − 1)d


Theorem 2. Using the Mellin transform, we get

w bt (e − 1) + ct (1 − e) − at + dt
MA (t) = (12)
a(1 − e) − b + c + (e − 1)d t
Proof.
 
∞ b
we[(x−a)/(b−a)]
MA (t) = xt−1 fA (x) dx = xt−1 dx
0 a a(1 − e) − b + c + (e − 1)d
 c  d
we we[(d−x)/(d−c)]
+ xt−1 dx + xt−1 dx
b a(1 − e ) − b + c + (e − 1 )d c a(1 − e ) − b + c + (e − 1)d
On integration, we obtain
  t
w ebt at w c bt
MA (t) = − + −
a(1 − e) − b + c + (e − 1)d t t a(1 − e) − b + c + (e − 1)d t t
 t
w d ect
+ −
a(1 − e) − b + c + (e − 1)d t t

t 
w b (e − 1) + ct (1 − e) − at + dt
=
a(1 − e) − b + c + (e − 1)d t


Theorem 3. The mean μA (x) and variance σA2 (x) of the random variable XA can be obtained as


w 1.72 b2 − 1.72 c2 − a2 + d2
μA (x) = (13)
a(1 − e) − b + c + (e − 1)d 2


w 1.72 b3 − 1.72 c3 − a3 + d3
σA2 (x) =
a(1 − e) − b + c + (e − 1)d 3


w 2
(1.72 b − 1.72 c2 − a2 + d2 )2
2
− (14)
[a(1 − e) − b + c + (e − 1)d]2 4
Proof.


w 1.72 b2 − 1.72 c2 − a2 + d2
μA (x) = e[XA ] = MA (2) =
a(1 − e) − b + c + (e − 1)d 2


w 1.72 b3 − 1.72 c3 − a3 + d3
σA2 (x) = MA (3) − [MA (2)]2 =
a(1 − e) − b + c + (e − 1)d 3


w2 (1.72 b2 − 1.72 c2 − a2 + d2 )2
−
[a(1 − e) − b + c + (e − 1)d]2 4

194 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198

Table 1
A comparison of the ranking results for different approaches.

Approaches Ex.1 Ex.2 Ex.3 Ex.4 Ex.5 Ex.6 Ex.7

Cheng [8] A<B AB Error AB A>B A<B<C Error

Chu [9] A<B AB Error A<B A>B A<B<C Error
Chen [5] A<B A<B A<B A<B A>B A<C<B A>B
Abbasbandy [1] Error AB A<B AB A<B A<B<C A>B
Chen [7] A<B A<B A<B A<B A>B A<B<C A>B
Kumar [13] A>B AB A<B A<B A>B A<B<C A>B
Pushpinder Singh [25] A<B A<B A<B A<B A>B A<B<C A>B
Rezvani [14] A<B A>B A>B AB AB A>C>B A<B
Proposed approach A>B A<B A<B A<B A<B A<C<B A>B

Definition 2. Let A = (a1 , b1 , c1 , d1 ; wA ) and B = (a2 , b2 , c2 , d2 ; wB ) are two generalized exponential fuzzy numbers, then

(i) If σA2 (x) > σB2 (x) then A > B,

(ii) If σA2 (x) < σB2 (x) then A < B,
(iii) If σA2 (x) ∼ σB2 (x) then A ∼ B .

3.1. Method to find the values of σA2 (x) and σB2 (x)

Let A = (a1 , b1 , c1 , d1 ; wA ) and B = (a2 , b2 , c2 , d2 ; wB ) are two generalized exponential fuzzy numbers, then use the following
steps to find the values of σA2 (x) and σB2 (x)
∗ Step 1. Find
 ∞  ∞
C μA (x) dx = 1, and C μB (x) dx = 1
−∞ −∞

∗ Step 2. Find μA (x) and μB (x)

∗ Step 3. Find σA2 (x) and σB2 (x)
∗ Step 4. Comparing A and B with use of Theorem 4.

4. Examples

In this section, seven sets of fuzzy numbers are compared using the proposed approach and existing approaches. The results
are shown in Table 1.

Example 1. Let A = (0.2, 0.4, 0.6, 0.8; 0.35) and B = (0.1, 0.2, 0.3, 0.4; 0.7) be two generalized trapezoidal fuzzy number, then
∗ step 1
2.72 2.72
CA = = = 2.21
0.2(1 − −2.72) − 0.4 + 0.6 + 0.8(2.72 − −1) 1.232
and
2.72 2.72
CB = = = 4.41
0.1(1 − −2.72) − 0.2 + 0.3 + 0.4(2.72 − −1) 0.616
∗ step 2

0.35 0.42 (2.72 − −1) + 0.62 (1 − −2.72) − 0.22 + 0.82
μA = = 0.0358
1.232 2
and

0.7 0.22 (2.72 − −1) + 0.32 (1 − −2.72) − 0.12 + 0.42
μB = = 0.0365
0.616 2
∗ step 3

σA2 (x) = 0.023 − (0.0358)2 = 0.0217

and

σB2 (x) = 0.011 − (0.03648)2 = 0.0097

∗ step 4
σA2 (x) > σB2 (x) then A > B .
 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198 195

Example 2. Let A = (0.1, 0.2, 0.4, 0.5; 1) and B = (0.1, 0.3, 0.3, 0.5; 1) be two generalized trapezoidal fuzzy number, then ∗ step 1
2.72 2.72
CA = = = 3.06
0.1(1 − −2.72) − 0.2 + 0.4 + 0.5(2.72 − −1) 0.888
and
2.72 2.72
CB = = = 3.95
0.1(1 − −2.72) − 0.3 + 0.3 + 0.5(2.72 − −1) 0.688
∗ step 2

1 0.22 (2.72 − −1) + 0.42 (1 − −2.72) − 0.12 + 0.52
μA = = 0.019
0.888 2
and

1 0.32 (2.72 − −1) + 0.32 (1 − −2.72) − 0.12 + 0.52
μB = = 0.174
0.688 2
∗ step 3
σA2 (x) = 0.0104 − (0.019)2 = 0.01
and
σB2 (x) = 0.06 − (0.174)2 = 0.03
∗ step 4
σA2 (x) < σB2 (x) then A < B .
Example 3. Let A = (0.1, 0.2, 0.4, 0.5; 1) and B = (1, 1, 1, 1; 1) be two generalized trapezoidal fuzzy number, then
2.72 2.72
CA = = = 3.06
0.1(1 − −2.72) − 0.2 + 0.4 + 0.5(2.72 − −1) 0.888
and
2.72
CB = =∞
(1 − −2.72) − 1 + 1 + 1(2.72 − −1)
∗ step 2

1 0.22 (2.72 − −1) + 0.42 (1 − −2.72) − 0.12 + 0.52
μA = = 0.019
0.888 2
and
μB = ∞
∗ step 3
σA2 (x) = 0.0104 − (0.019)2 = 0.01
and
σB2 (x) = ∞
∗ step 4
σA2 (x) < σB2 (x) then A < B .
Example 4. Let A = ( − 0.5, −0.3, −0.3, −0.1; 1) and B = (0.1, 0.3, 0.3, 0.5; 1) be two generalized trapezoidal fuzzy number, then
∗ step 1
2.72 2.72
CA = = = 3.95
(−0.5)(1 − −2.72) + 0.3 − −0.3 − 0.1(2.72 − −1) 0.688
and
2.72 2.72
CB = = = 3.95
0.1(1 − −2.72) − 0.3 + 0.3 + 0.5(2.72 − −1) 0.688
∗ step 2

1 (−0.3)2 (2.72 − −1) + (−0.3)2 (1 − −2.72) − (−0.5)2 + (−0.1)2
μA = = −0.174
0.688 2
and

1 0.32 (2.72 − −1) + 0.32 (1 − −2.72) − 0.12 + 0.52
μB = = 0.174
0.688 2
196 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198

∗ step 3
σA2 (x) = −0.06 − (−0.174)2 = −0.01
and
σB2 (x) = 0.06 − (0.174)2 = 0.03
∗ step 4
σA2 (x) < σB2 (x) then A < B .
Example 5. Let A = (0.3, 0.5, 0.5, 1; 1) and B = (0.1, 0.6, 0.6, 0.8; 1) be two generalized trapezoidal fuzzy number, then ∗ step 1
2.72 2.72
CA = = = 9.1
0.3(1 − −2.72) − 0.5 + 0.5 + (2.72 − −1) 0.3
and
2.72 2.72
CB = = = 2.26
0.1(1 − −2.72) − 0.6 + 0.6 + 0.8(2.72 − −1) 1.204
∗ step 2

1 0.52 (2.72 − −1) + 0.52 (1 − −2.72) − 0.32 + 12
μA = = 1.51
0.3 2
and

1 0.62 (2.72 − −1) + 0.62 (1 − −2.72) − 0.12 + 0.82
μB = = 0.26
1.204 2
∗ step 3
σA2 (x) = 1.1 − (1.51)2 = −1.2
and
σB2 (x) = 0.14 − (0.26)2 = 0.1
∗ step 4
σA2 (x) < σB2 (x) then A < B .
Example 6. Let A = (0, 0.4, 0.6, 0.8; 1) and B = (0.2, 0.5, 0.5, 0.9; 1) and C = (0.1, 0.6, 0.7, 0.8; 1) be three generalized trapezoidal
fuzzy number, then ∗ step 1
2.72 2.72
CA = = = 1.73
0(1 − −2.72) − 0.4 + 0.6 + 0.8(2.72 − −1) 1.576
and
2.72 2.72
CB = = = 2.26
0.2(1 − −2.72) − 0.5 + 0.5 + 0.9(2.72 − −1) 1.204
and
2.72 2.72
CC = = = 2.1
0.1(1 − −2.72) − 0.6 + 0.7 + 0.8(2.72 − −1) 1.304
∗ step 2

1 0.42 (2.72 − −1) + 0.62 (1 − −2.72) − 0 + 0.82
μA = = 0.09
1.576 2
and

1 0.52 (2.72 − −1) + 0.52 (1 − −2.72) − 0.22 + 0.92
μB = = 0.32
1.204 2
and

1 0.62 (2.72 − −1) + 0.72 (1 − −2.72) − 0.12 + 0.82
μC = = 0.16
1.304 2
∗ step 3
σA2 (x) = 0.05 − (0.09)2 = 0.0419
and
σB2 (x) = 0.2 − (0.32)2 = 0.0976
 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198 197

and
σC2 (x) = 0.07 − (0.16)2 = 0.0444
∗ step 4
σA2 (x) < σC2 (x) < σB2 (x) then A < C < B .
Example 7. Let A = (0.1, 0.2, 0.4, 0.5; 1) and B = ( − 2, 0, 0, 2; 1) be two generalized trapezoidal fuzzy number, then ∗ step 1
2.72 2.72
CA = = = 3.06
0.1(1 − −2.72) − 0.2 + 0.4 + 0.5(2.72 − −1) 0.888
and
2.72 2.72
CB = = = 0.39
(−2)(1 − −2.72) − 0 + 0 + 2(2.72 − −1) 6.88
∗ step 2

1 0.22 (2.72 − −1) + 0.42 (1 − −2.72) − 0.12 + 0.52
μA = = 0.019
0.888 2
and

1 0(2.72 − −1) + 0(1 − −2.72) − (−2)2 + 22
μB = =0
6.88 2
∗ step 3
σA2 (x) = 0.0104 − (0.019)2 = 0.01
and
σB2 (x) = 0
∗ step 4
σA2 (x) > σB2 (x) then A > B . For the validation, the results of the approach are compared with different existing approaches in
Table 1.

5. Conclusion

As we know that any probability density function with finite support is associated with an expected value. In this method, we
define the representation of generalized exponential fuzzy number based on the integral value of graded mean h-level and values
are calculated by finding expected values using the probability density function corresponding to the membership functions
of the given fuzzy number. The main advantage of the proposed approach is that the proposed approach provides the correct
ordering of generalized and normal trapezoidal fuzzy numbers and also the proposed approach is very simple and easy to apply
in the real life problems.

References

[1] S. Abbasbandy, T. Hajjari, A new approach for ranking of trapezoidal fuzzy numbers, Comput. Math. Appl 57 (2009) 413–419.
[2] G. Bortolan, R. Degani, A review of some methods for ranking fuzzy subsets, Fuzzy Sets Syst. 15 (1) (1985) 119.
[3] S.K. Barik, M.P. Biswal. Probabilistic Quadratic Programming Problems with Some Fuzzy Parameters, Hindawi Publishing Corporation, Advances in Operations
Research, vol. 2012, Article ID 635282, 13 pp.
[4] S.-H. Chen, Ranking fuzzy numbers with maximizing set and minimizing set, Fuzzy Sets Syst. 17 (2) (1985) 113–129.
[5] S.j. Chen, S.M. Chen, Fuzzy risk analysis based on the ranking of generalized trapezoidal fuzzy numbers, Appl. Intell. 26 (2007) 1–11.
[6] S.-H. Chen, G.-C. Li, Representation, ranking, ang distance of fuzzy number with exponential membership function using graded mean integration method,
Tamsui Oxford J. Math. Sci 16 (2000) 123–131.
[7] S.M. Chen, J.H. Chen, Fuzzy risk analysis based on ranking generalized fuzzy numbers with different heights and different spreads, Expert Syst. Appl. 36
(2009) 6833–6842.
[8] C. Cheng, A new approach for ranking fuzzy numbers by distance method, Fuzzy Sets Syst. 95 (3) (1998) 307–317.
[9] T. Chu, C. Tsao, Ranking fuzzy numbers with an area between the centroid point and original point, Comput. Math. Appl. 43 (1–2) (2002) 111–117.
[10] D. Dubois, H. Prade, The mean value of a fuzzy number, Fuzzy Sets Syst. 24 (3) (1987) 279–300.
[11] R. Jain, Decision making in the presence of fuzzy variables, IEEE Trans. Syst. Man Cybernet. 6 (10) (1976) 698–703.
[12] C. Liang, J. Wu, J. Zhang. Ranking indices and rules for fuzzy numbers based on gravity center point, Paper presented at the 6th world Congress on Intelligent
Control and Automation, Dalian, China, 2006, pp. 21–23.
[13] A. Kumar, P. Singh, A. Kaur, P. Kaur, Rm approach for ranking of generalized trapezoidal fuzzy numbers, Fuzzy Inf. Eng. 2 (1) (2010) 37–47.
[14] S. Rezvani, Graded mean representation method with triangular fuzzy number, World Appl. Sci. J. 11 (7) (2010) 871–876.
[15] S. Rezvani, Multiplication operation on trapezoidal fuzzy numbers, J. Phys.Sci. 15 (2011) 17–26.
[16] S. Rezvani, A new method for ranking in perimeters of two generalized trapezoidal fuzzy numbers, Int. J Appl. Oper. Res. 2 (3) (2012) 83–90.
[17] S. Rezvani, A new approach ranking of exponential trapezoidal fuzzy numbers, J. Phys. Sci. 16 (2012) 45–57.
[18] S. Rezvani, Ranking method of trapezoidal intuitionistic fuzzy numbers, Ann. Fuzzy Math. Inf. 5 (3) (May 2013) 515–523.
[19] S. Rezvani, Ranking generalized trapezoidal fuzzy numbers with euclidean distance by the incentre of centroids, Math. Aeterna. 3 (2) (2013) 103–114.
[20] S. Rezvani, Representation of trapezoidal fuzzy numbers with shape function, Ann. Fuzzy Math. Inform. 8 (1) (2014) 89–112.
[21] S. Rezvani, A new method for ranking fuzzy numbers with using trd distance based on mean and standard deviation, Int. J. Mech. Electr. Comput. Technol.
4 (12) (2014) 840–856.
198 S. Rezvani / Applied Mathematics and Computation 262 (2015) 191–198

[22] S. Rezvani, Representation of generalized triangular fuzzy sets, Int. J. Phys. Math. Sci. 5 (1) (2015) 41–52.
[23] S. Rezvani, A new similarity measure of generalized fuzzy numbers based on left and right apex angles (i), Palestine J. Math. 4 (1) (2015) 117–126.
[24] R. Saneifard, R. Saneifard, A modified method for defuzzification by probability density function, J. Appl. Sci. Res. 7 (2) (2011) 102–110.
[25] Pushpinder Singh, Ranking of generalized trapezoidal fuzzy numbers based on rank, mode, divergence and spread, Turkish J. Fuzzy Syst. (eISSN: 1309û1190)
1 (2) (2010) 141–152.
[26] Y.J. Wang, H.S. Lee, The revised method of ranking fuzzy numbers with an area between the centroid and original points, Comput. Math. Appl. 55 (2008)
2033–2042.
[27] L.A. Zadeh, Fuzzy Set, Inf. Control 8 (3) (1965) 338–353.