Aggregation of Opinions for System Selection Using Approximations of Fuzzy Numbers

Applied Mathematics and Sciences: An International Journal (MathSJ ), Vol. 1, No. 2, August 2014
75
AGGREGATION OF OPINIONS FOR SYSTEM
SELECTION USING APPROXIMATIONS OF
FUZZY NUMBERS
D. Stephen Dinagar1
, K.Jivagan2
,
1,2
PG and Research Department of Mathematics, T.B.M.L. College, Porayar
ABSTRACT
In this article we assume that experts express their view points by way of approximation of Triangular
fuzzy numbers. We take the help of fuzzy set theory concept to model the situation and present a method to
aggregate these approximations of triangular fuzzy numbers to obtain an overall approximation of
triangular fuzzy number for each system and then linear ordering done before the best system is chosen. A
comparison has been made betweenapproximation of triangular fuzzy systems and the corresponding fuzzy
triangular numbers systems. The notions like fuzziness and ambiguity for the approximation of triangular
fuzzy numbers are also found.
Keywords :
Approximations ofTriangular fuzzy numbers, similarity of fuzzy numbers, Relative similarity degree.
1. INTRODUCTION
In any decision making problem, when one has to select froma finite number of systems, opinions
of different decision makers are sought. Each expert/decision makers has his own way of
assessing a system and thus provides his own rating or grading for that system. Each expert may
prefer to express their view point in an imprecise manner rather than an exact manner. It is
because of this imprecision or vagueness inherent in the subject assessment of different decision
makers/experts, that the help of fuzzy set theory is sought. Once opinions are expressed by the
decision makers, the question arises, how best to aggregate these individual opinions into a
general consensus opinion. Aggregation operations on fuzzy sets are operations by which several
fuzzy sets are combined to produce a single fuzzy set. Different ways of aggregating opinions
have been suggested by many works. Nurmi, (1981) [7], Tanino (1984) [8], Fedrizzi, and
kacprazyk (1988) [3] proposed that each expert assigns a fuzzy preference relation and these
individual fuzzy preference relations were then aggregated into a group fuzzy preference relation
in order to determine the best alternative. Bardossy et al. (1993) [2] proposed five aggregation
techniques, namely crisp weighting, fuzzy weighting, minimal fuzzy extension, convex fuzzy
extension and mixed linear extension method. Hsu and Chen (1996)[4] suggested a method of
aggregation by which a consensus opinion is arrived at, on evaluating positive trapezoidal fuzzy
numbers that represent an individual’s subjective estimate. They employed the method called
Similarity Aggregation Method (SAM) to find out an agreement between the experts. In one of
our recent works, we have suggested some refinement in the procedure given by Hsu and Chen
(1996) [4] which iscomputationally simpler and is easy to understand.
Once aggregated opinions for each system is obtained, a selection of the best system has to be
made.

76
2. PRELLIMINARIES
2.1. FUZZY SETS
Let X denote a universal set i.e., x={x}; then the characteristic function which assigns certain
values or a membership grade to the elements of this universal set within a specified range {0, 1}
is known as the membership function and the set thus defined is called a fuzzy set. The
membership graders correspond to the degree to which an element is compatible with the concept
represented by the fuzzy set
If µÃis the membership function definingÃ fuzzy set Ã, then,
µÃ: X → [0, 1]
Where [0, 1] developed the interval of real numbers from 0 to 1.
2.2 α
α
α
α-CUT
An α-cut of a fuzzy set Ã is a crisp set Ãα that contains all the elements of the universal set X that
have a membership grade in A greater than or equal the specified value of α. Thus,
Thus, Ã = {x∈X; Ã
µ 1
0
,
)
( ≤
≤
≥ x
x α }
2.3 FUZZY NUMBER
A fuzzy subject Ãof the real line R with membership function Ã
µ : X → [0, 1] is called a fuzzy
number if,
a) Ã is normal, i.e., there exist an element x0 such that µÃ (x0) = 1.
b) Ã is fuzzy convex, i.e.,µÃ (λx1 + (1-λ) x2) ≥ Ã
µ (x1) ∧ Ã
µ (x2) ∀x1, x2 ∈R.
c) µÃ is a upper semi continuous and
d) Sup Ã is bounded, where sup Ã = { }
0
(x)
; ≥
∈ Ã
R
x µ
2.4 POSITIVE FUZZY NUMBER
A fuzzy number A is called positive fuzzy number if its member ship function is such that µÃ(x)
= 0, ∀x < 0.
This is denoted by Ã> 0.
2.5 TRIANGULAR FUZZY NUMBER
A triangular fuzzy number Ã is denoted as Ã = (a1, a2, a3) and is defined by the membership
function as,

77











>
≤
≤
−
≤
≤
−
<
=
3
3
2
2
3
3
2
1
1
2
1
1
Ã
if
0
a
a
-
a
a
a
a
-
a
if
0
(x)
a
x
a
x
if
x
a
x
if
a
x
a
x
µ
It can be characterized by defining the interval of confidence at levelα. Thus for all α∈[0, 1]
Ãα = [(a2-a1) α+a1, a3-(a3-a2)α]
2.6 APPROXIMATION OF TRIANGULAR FUZZY NUMBER
Let Ã = (a1, a2, a3) be a triangular fuzzy number and the approximation of triangular fuzzy number
is defined as Ã* = (t1, t2, t3)
Where t1= aL= inf {x∈A/µÃ (x) ≥0.5} =
2
2
1 a
a +
t2= a2 = {x∈A/µÃ (x) =1}
t3= aU= sup {x∈A/µÃ (x) ≥0.5}=
2
3
2 a
a +
1.0
0.5
a1aL a2aUa3
2.7 AMBIGUITY OF FUZZY NUMBER [9]
Let Ã be a fuzzy number with α-cut representation (AL(α)), AU(α), then the ambiguity of Ã is
defined as,
Amb(Ã)= dx
A
A L
U )]
(
)
(
[
1
0
α
α
α −
∫
2.8 FUZZINESS OF FUZZY NUMBER
Let Ã be a fuzzy number with α-cut representation (AL(α)), AU(α), then the fuzziness of Ã is
defined as,

78
Fuzz (Ã)= dx
A
A L
U )]
(
)
(
[
2
1
0
α
α −
∫ + dx
A
A U
L )]
(
)
(
[
1
2
1
α
α −
∫
2.9 ASSOCIATED APPROXIMATION OF TRIANGULAR FUZZY NUMBER
If Ã*= (t1, t2, t3) is an approximation of Triangular fuzzy number, then its associated value is
given by
4
t
2t
t
* 3
2
1 +
+
=
Α
∧
3. AGGREGATION OF OPINIONS
3.1 NOTATIONS
µÃ(x) – Membership function for the element with respect to the fuzzy subset A.
Ãi* – Approximation of Triangular fuzzy number corresponding to system Ãi.
Ãiα* – α-cut of the approximation of Triangular fuzzy number Ãi or the interval
of confidence at level α∈[0, 1] of the approximation of Triangular fuzzy
number Ãi*.
LD (Ãi*, Ãj*) – Left distance between two approximation of Triangular fuzzy number Ãi*
and Ãj*.
∂(Ãi*, Ãj*) – Normalized distance between two approximation of Triangular fuzzy
number Ãi* and Ãj*.
D (Ãi*, Ãj*) – level of similarity between two triangular fuzzy number Ãi* and Ãj*.
Ei - ith
expert / decision maker
A(Ei) – Average similarity degree of expert Ei(i=1, . . . n).
RDSi – Relative similarity degree of expert Ei(i=1, . . . n).
Ã* – Overall Approximation of Triangular fuzzy number.
Pi – weights given to the expert Ei(i=1, . . . n) according to his importance.
Wi – Relative degree of importance of the expert Ei(i=1, . . . n)
CDCi – consensus similarity degree coefficient of the expert Ei(i=1, . . . n)
γ - Weights attached to the relative degree of importance of the expert Ei(i=1,
. . . n)
(1-γ) –Weight attached to the relative, similarity degree of the expert Ei(i=1, . . . n)
(β1-β2) – Arbitrary values at the left and right respectively of triangular fuzzy
number, chosen such that
(β2-β1) ≥
2
)]
Ã
,
Ã
(
)
Ã
,
Ã
(
[ j
i
j
i RD
LD +
(.) – Product operator of fuzzy numbers.
3.2 ALGORITHMS
3.2.1 Aggregation of opinions in terms of approximations of triangular fuzzy is to select the
best system when each expert is equally important.
Step 1:
A finite set of experts / decision makers Ei(i=1, . . . n) give their subjective estimates

79
of the alternatives R, P, Q,M, . . . in terms of approximation of triangular fuzzy numbers
~
~
~
~
*
,
*
,
*
,
* i
i
i M
Q
P
R i
.write down these numbers in terms of intervals of confidence
........
,
*
,
*
,
*
,
*
~
~
~
~
α
α
α
α i
i
i M
Q
P
R i
Step 2:
Calculate normalized distance
∂ )
*
,
*
(
~
~
j
i
R
R , ∂( i
P*
~
, j
P*
~
), ∂( i
Q*
~
, j
Q*
~
), ∂( i
M *
~
, j
M*
~
), . . . . between every pair of
approximation of Triangular fuzzy numbers j
i
R
R *
,
*
~
~
(for system R), j
i
P
P *
,
*
~
~
(for system P),
j
i
Q
Q *
,
*
~
~
(for system Q), j
i
M
M *
,
*
~
~
(for system M) and so on (i, j = 1 . . . .n) where, for
approximation of Triangular fuzzy numbers i
A*
~
.
∂( j
i
A
A *
*
~
~
− ) = 1
0
)],
*
*,
(
)
*
*,
(
[
)
(
2
1 ~
~
~
~
1
2
<
∂
≤
+
−
j
i
j
i A
A
RD
A
A
LD
β
β
(kaufmann et al. (1985)[ ])
If the interval of confidence of approximation of triangular fuzzy number i
A*
~
and j
A*
~
be, respectively,
[ ]
α
α
α
)
(
,
)
(
* 2
3
3
1
1
2
~
a
a
a
a
a
a
A i
−
−
+
−
=
[ ]
α
α
α
)
(
,
)
(
* 2
3
3
1
1
2
~
b
b
b
b
b
b
A i
−
−
+
−
=
Then
LD ( i
A*
~
, j
A*
~
) = [ ]
1
1
2
1
1
2 )
(
)
( b
b
b
a
a
a −
−
−
+
− α
α
And
RD ( i
A*
~
, j
A*
~
) = [ ]
3
2
3
3
2
3 )
(
)
( b
b
b
a
a
a −
−
+
+
−
− α
α
Step 3:
Calculate the levels of similarity D( ,
*
~
i
R j
R*
~
), D( ,
*
~
i
P j
P*
~
), D( ,
*
~
i
Q j
Q*
~
), D( ,
*
~
i
M j
M *
~
)
.
For approximation of triangular fuzzy numbers ( i
*
~
Α , j
*
~
Α )
D ( i
A*
~
, j
A*
~
) = 1 - ∂( i
*
~
Α , j
*
~
Α )
= 1- 1
0
)
(
2
)]
*
,
*
(
)
*
,
*
(
[
1
2
~
~
~
~
<
∂
≤
−
Α
Α
+
Α
Α
β
β
j
i
j
i
RD
LD

80
Step 4:
Construct a similarity matrix SM where E1 . . . . .E2. . . . .Ei. . . . .Ej. . . . .En
SM =
n
i
Ε
Ε
Ε
Ε
.
.
.
.
2
1




















1
......
.......
......
.......
1
......
.......
1
......
.......
......
1
2
1
2
.
.
1
2
2
2
.
.
21
1
1
1
12
j
n
n
i
n
n
n
i
j
i
i
i
n
j
i
n
j
i
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
D
Where Dii=1 ∀i=1 . . . .n and Dij=D( i
*
~
Α , j
*
~
Α ) ∀i ≠1
It is to be noted that SM is a symmetric matrix.
Step5:
Calculate average similarity degree of expert Ei(i=1, . . . . . n) as
A(Ei) = ∑
≠
=
−
n
j
j
ij
D
n
1
1
1
1
Step 6:
Calculate the relative similarity degree RSDi as
RSDi=
∑
=
n
j
i
i
E
A
E
A
1
)
(
)
(
Step 7:
Overall fuzzy number or aggregated consensus opinion is given by
=
~
A ∑
=
n
i
i
RSD
1
(.)
(
i
*
~
Α )
Step 8:
Find out the associated ordinary number corresponding to each overall Approximation of fuzzy
number, where the
=
Α*
4
2 3
2
1 t
t
t +
+
Step 9:
Arrange them in descending order. The one coming at the 1st
place is best alternative and is
selected.

81
3.2.2 Aggregation of opinions of experts and selection of the best alternative when some
experts are more important than others.
Let there be some experts who are more important than the others. Let these experts be given a
weight Pi=1. Then the steps of this procedure are as follows.
Step 1:
Find out the relative importance weight Pi of all other experts with respect to most
Important experts.
Step 2:
Calculate the relative degree of importance Wi of each experts as
Wi = ∑
∑ =
=
=
n
i
i
n
i
i
i
W
P
P
1
1
1
,
Step 3:
Repeat steps 1 to 6 of case I
Step 4:
Calculate consensus similarity degree coefficient as
CDCi = γwi + (1- γ)RSDi, 0 ≤γ≤ 1
Step 5:
The overall Approximationof fuzzy number of the combination of experts opinion is then given
by
=
~
A ∑
=
n
i
i
i A
CDC
1
~
)
(.)
(
Step 6:
Repeat steps 8 and 9 for selection of the best alternative CDCi is convex combination of weights
attached to an expert and his relative similarity degree. The weights attached to them areγand (1-
γ) respectively. One has to make a judicious choice ofγ, keeping in mind clearly. The importance
of relative similarity degree and weight of the expert. Further if γ = 0, then the determining factor
is only relative similarity degree which is the case when all the experts are equally important (w1
= w2 = . . . . . . . .wn=1/n) where as if γ=1, full weight is attached to the relative degree of
importance wi and the relative similarity degree RSDi becomes irrelevant. For the purpose of
illustration we take the same value of γ as has been used in Hsu and Chen (1996) [4].

82
4. Illustrations
(Case I)
Let us consider an illustration from [1]. Let there be three experts Ε1,Ε2 andΕ3 and four
alternatives/systems, R, P, Q, and M. Let the experts give their opinion an each alternatives
interms of Triangular fuzzy numbers as follows,
System
Expert
R P Q M
E1 (2,4,5) (1,2.5,4) (4,5,6) (3.5,5.5,6.5)
E2 (2.5,4,6) (1.5,3,5) (3,5,8) (2,6,8)
E3 (3,5,7) (2,3.25,6) (2,4,5) (1,3,6)
Then the approximation of the above corresponding triangular fuzzy numbers as follows
System
Expert
R* P* Q* M*
E1 (3,4,4.5)
(1.75, 2.5,
3.25)
(4.5, 5, 5.5) (4.5, 5.5, 6)
E2 (3.25, 4.5) (2.25, 3,4) (4,5,6.5) (4,6,7)
E3 (4, 5, 6)
(2.62, 3.25,
4.62)
(3, 4, 4.5) (2, 3, 5.5)
Out of the four alternations, the best alternative/system has to be chosen. Expressing the
approximation of triangular fuzzy numbers in terms of confidence at level α, α∈[0,1], we have
s∀
)
)
5
.
0
(
5
.
4
,
3
(
~
1
*
α
α
α −
+
=
R ; )
5
)
75
.
0
(
25
.
3
(
*2
~
α
α −
+
=
R ; )
6
,
4
(
*3
~
α
α
α −
+
=
R
)
)
5
.
6
5
.
5
,
)
5
.
0
(
5
.
4
(
*1
~
α
α
α −
+
=
Q ; )
)
5
.
1
(
5
.
6
,
4
(
~
2
*
α
α
α −
+
=
Q ; )
)
5
.
6
(
5
.
4
,
3
(
*3
~
α
α
α −
+
=
Q
)
)
75
.
0
(
25
.
3
,
)
75
.
0
(
75
.
1
(
*
1
~
α
α
α −
+
=
P ; )
4
,
)
75
.
0
(
25
.
2
(
*2
~
α
α
α −
+
=
P ; )
)
32
.
1
(
62
.
4
,
)
63
.
0
(
65
.
2
(
*3
~
α
α
α −
+
=
P
)
)
5
.
6
(
6
,
5
.
4
(
*1
~
α
α
α −
+
=
M ; )
)
5
.
6
(
6
,
5
.
4
(
*2
~
α
α
α −
+
=
Μ ; )
)
5
.
2
(
5
.
5
,
2
(
*3
~
α
α
α −
+
=
Μ
Since distance are always positive, whenever, we get a negative value it will be taken as positive.
Now we calculate normalized distances between every pair of experts and for each system.
Using the formulas given in kaufmann and Gupta (1985) [6] (using α=0.5) δ ( i
R
~
* , j
R
~
* ),δ (
j
i *
,
*
~
~
Ρ
Ρ ),δ ( j
i Q
Q
~
~
, ), and δ ( j
i
~
~
, Μ
Μ ) can easily be computed because in ATFN’s, we have
only straight lines.
For system *
R ,we take the arbitrary values ofβ1as 3 (minimum value of the ATFN) and β2 as 6
(Maximum value of ATFN).
LD ( 1
~
*
R , 2
~
*
R )= 3+α-3.25-(0.75)α=-0.25+(0.25) α

83
RD ( 1
~
*
R , 2
~
*
R ) = 4.5-(0.5) α-5+α = -0.5+(0.5) α
∂12 = 0625
.
0
6
)
75
.
0
(
75
.
0
)
3
6
(
2
5
.
0
)
5
.
0
(
)
25
.
0
(
25
.
0
=
+
−
=
−
−
+
+
− α
α
α
And D12=1-∂12=1-0.0625=0.9375
∂13 = 0.456 ∂23 = 0.3125
∂13 = 0.542 ∂23 = 0.6875
Hence, the similarity Matrix can be written as
E1 E2 E3
SM=
3
2
1
E
E
E










1
687
.
0
542
.
0
687
.
0
1
937
.
0
542
.
0
937
.
0
1
The average similarity degree is calculated as
A (E1) = ½ [0.937+0.542] =0.7395
A (E2) = 0.812
A (E3) = 0.614
Then the relative similarity degree is given by
RSD1=
∑
=
3
1
)
(
)
(
i
i
i
E
A
E
A =
16
.
2
7395
.
0
= 0.3423
RSD2= 0.3759
RSD3= 0.2842
So the overall fuzzy number or aggregated fuzzy opinion is
*
~
R = 0.3453 (3, 4, 4.5) +0.3759 (3.25, 4, 5) +0.2842(4, 5, 6)
= (3.3853, 4.2938, 5.125)
And
]
)
8312
.
0
(
125
.
5
,
)
9085
.
0
(
3853
.
3
[
*
~
α
α
α −
+
=
R
The associated ordinary number corresponding to this overall approximation fuzzy number is

84
*
∧
R = 0.3428(3, 4, 4.5) +0.3759(3.25, 4.5) +0.2842(4, 5, 6)
= (3.3853, 4.2938, 5.125)
And
^
~
]
)
8312
.
0
(
125
.
5
,
)
9085
.
0
(
3853
.
3
[
* α
α
α −
+
=
R
The associated ordinary number corresponding to this overall approximation of triangular
fuzzy number is
*
∧
R =
4
9896
.
5
)
2938
.
4
(
2
3853
.
3 +
+
*
∧
R = 4.4906
For system P*, we have β1=1.75 and β2 = 4.62
∂12 = 0.1959; D12 = 0.8041
∂13 = 0.325; D13 = 0.6742
∂23 = 0.0540; D23 = 0.946
Hence the similarity matrix is
SM =










1
9542
.
0
6742
.
0
9542
.
0
1
8041
.
0
6742
.
0
8041
.
0
1
By using the algorithm,
The overall fuzzy number is given as
~
Ρ *=0.3041 (1.75, 2.5, 3.25) +0.3617(2.25, 3, 4) +0.3350 (2.62, 3.25, 4.62)
The associated ordinary number is given by
∧
Ρ *=2.9014
For system Q*, we have β1=3 and β2 = 6.5.
∂12 = 0.1071; D12 = 0.8928
∂13 = 0.1785; D13 = 0.8215
∂23 = 0.2142; D23 = 0.7858
SM =










1
7858
.
0
8215
.
0
7858
.
0
1
8928
.
0
8215
.
0
8928
.
0
1

85
~
Q*= 0.3428 (4.5, 5, 5.5) +0.3357(4, 5, 6.5) +0.3214(3, 4, 4.5)
= (3.8496, 4.6781, 5.5137)
∧
Q* = 4.6798
For system M*, we have β1=2 and β2 = 7
∂12 = 0.075; D12 = 0.925
∂13 = 0.4; D13 = 0.6
∂23 = 0.475; D23 = 0.525
SM =










1
525
.
0
6
.
0
525
.
0
1
925
.
0
6
.
0
925
.
0
1
~
M *=0.3719(4.5, 5.5, 6) +0.3536(4, 6, 7) +0.2741(2, 3, 5.5)
= (3.6361, 4.9893, 6.2141)
*
∧
M = 4.9572
Collecting all the associated numbers, one each for each system, we have
^
R*= 4.4906
^
Q*= 4.6798
^
R
^
Ρ *= 2.9014
^
M *= 4.9572
Ordering them linearly in decreasing order i.e., the one having the maximum value is placed first,
we have
∧
Μ *>
∧
R *>
∧
Q*>
∧
P *
Hence the system
∧
Μ *is chosen. If the system M* is somehow not available, then the next one
i.e., Q* is chosen, and so on.

86
1.0-
0.8-
0.6-
0.4-
0.2-
1 2 3 4 5 6
Case (II)
Let us consider the same illustration as in case I. Out of the 3 experts E1, E2 and E3.Let the expert
E1 be most important. Hence we give a weight P1=1 to him.
The importance of other experts relative to him are say for E2, P3=0.6 and for E3, β3=0.2
Then the relative degrees of importance are
W1 =
2
.
0
6
.
0
1
1
+
+
= 0.5555 = 0.56
W2 = 0.33; W3 = 0.11,s.t ∑
=
=
3
1
1
i
i
w
Considering the values of RSDi’s from illustration I case I, for system R*, the consensus degree
coefficient for experts E1, E2 and E3 are, respectively.
CDC1 = 0.4 x 0.56 + 0.6 x 0.3423 = 0.4293
CDC2 = 0.4 x 0.33 + 0.6 x 0.3759 = 0.3575
CDC3 = 0.2145
Where γ = 0.4 (as w1> RSD1)
Thus, the overall fuzzy number is given us
~
R *= 0.4293(3, 4, 5,) +0.3575(3.25, 4, 5) +0.2145(4, 5, 6)
= (3.3075, 4.2197, 5.0063)
And
α
~
R *= (3.3075+ (0.9122)α, 5.0063-(0.7866)α)
Therefore the associated ordinary number corresponding to this triangular fuzzy number is
*
∧
Μ
*
∧
Q
*
∧
R
*
∧
Ρ

87
~
R *=
4
2 3
2
1 a
a
a +
+
=
4
0063
.
5
)
2197
.
4
(
2
3075
.
3 +
+
∧
R *=
4
4394
.
8
3138
.
8 +
= 4.1883
Similarly for system P*, we have
The associated ordinary number corresponding to this triangular fuzzy number is
∧
Ρ *= 2.2963
Similarly for system Q*, we have
The associated ordinary number corresponding to this number is
=
∧
*
Q 4.7742
Similarly for the system M*, we have
The associated ordinary number is
=
Μ
∧
* 5.0850
Collecting all the associated ordinary numbers of overall fuzzy numbers, one each for each
system, we have
=
*
^
R 4.1883; =
*
^
Q 4.7742
=
Ρ *
^
2.2963; =
*
^
M 5.0850
Ordering them in descending order, we get
∧
Μ *>
∧
Q*>
∧
R *>
∧
P *
Hence, the system M* is to be chosen. If system M* is not available, system Q* is chosen and so
an.
1.0-
*
∧
Μ
*
*
∧
Q
*
∧
R
*
∧
Ρ

88
0.8-
0.6-
0.4-
0.2-
1 2 3 4 5 6
5. Comparison between fuzzy numbers and its approximation of fuzzy
numbers
Case (i)
System TFN ATFN
R* Amb(ܴ
෨) = 0.5824
Fuzz(ܴ
෨) = 0.8737
Amb(ܴ
෨∗
) = 0.29
Fuzzy(ܴ
෨∗
)=
0.124
P* Amb(ܲ
෨) = 0.5836
Fuzz(ܲ
෨) = 0.876
Amb(ܲ
෨∗
) =
0.2150
Fuzz(ܲ
෨∗
)=
0.3226
Q* Amb(ܳ
෨) = 0.5553
Fuzz(ܳ
෨)= 0.8335
Amb(ܳ
෨∗
)=
0.2773
Fuzz(ܳ
෨∗
)=
0.4160
M* Amb(‫ܯ‬
෩)= 0.7729
Fuzz(‫ܯ‬
෩) = 1.1594
Amb(‫ܯ‬
෩∗
)=
0.4296
Fuzz(‫ܯ‬
෩∗
) =
0.6445
Table. 5a
Case (ii)
System TFN ATFN
R* Amb(ܴ
෨) = 0.5678 Amb(ܴ
෨∗
) = 0.2831

89
Fuzz(ܴ
෨) = 0.8517 Fuzzy(ܴ
෨∗
) = 0.4247
P* Amb(ܲ
෨) = 0.5685
Fuzz(ܲ
෨) = 0.8528
Amb(ܲ
෨∗
) = 0.2850
Fuzz(ܲ
෨∗
) = 0.4276
Q* Amb(ܳ
෨) = 0.5413
Fuzz(ܳ
෨) = 0.5120
Amb(ܳ
෨∗
) = 0.2697
Fuzz(ܳ
෨∗
) = 0.4045
M* Amb(‫ܯ‬
෩) = 0.7460
Fuzz(‫ܯ‬
෩) = 1.1190
Amb(‫ܯ‬
෩∗
)= = 0.4054
Fuzz(‫ܯ‬
෩∗
)= = 0.6081
Table. 5b
6. CONCLUSION
A comparison has been made between approximation of triangular fuzzy number systems and the
corresponding fuzzy triangular numbers systems, with the aid of notions like fuzziness and
ambiguity for the approximation of fuzzy numbers. It can also be seen that from the section V,the
fuzziness and ambiguity of theapproximationof triangular fuzzy number system is very less than
the triangular fuzzy number system which would be an important point to be noted for the future
works on the approximations of fuzzy numbers.
REFERENCES
[1] Ashok Kumar,Aggregation of Opinions for System Selection Under Fuzzy Environment., Recent
Development in operational Research, ManjuLata and KanwarSen (Eds) (2001) Narosa Publishing
House,New Delhi, India.
[2] Bardossy,A., Ducktein, L., Bogardi, I (1993). Fuzzy sets and systems, 57,173-181.
[3] Fedrizzi, M., and Kacprzyk,J.(1988). Non-conventional Preference Relation in Decision
Making(Berlin:Springer).
[4] Hsu, H.M., and Chen, C.T. (1996). Fuzzy Sets and Systems, 79, 279-285.
[5] Kacprzyk, J., and Fedrizzi, M. (1988). European J. Oper, Res., 34, 315-325.
[6] Kaufmann, A., and Gupta, M.M., (1985). Introduction to Fuzzy Arithmetic, Theory and
Applications. (New York: Van-Nostrand Reinhold).
[7] Nurmi, H. (1981). Fuzzy Sets and Systems, 6, 249-259.
[8] Tanini, T. (1984). Fuzzy Sets and systems, 12, 117-131.
[9] Stephen Dinagar .D, Jivagan .K, A note on interval approximation of fuzzy numbers, proceedings of
the international conference on mathematical methods and computation, ICOMAC, 2014.

Aggregation of Opinions for System Selection Using Approximations of Fuzzy Numbers

More Related Content

Similar to Aggregation of Opinions for System Selection Using Approximations of Fuzzy Numbers (20)

More from mathsjournal (20)

Recently uploaded (20)

Aggregation of Opinions for System Selection Using Approximations of Fuzzy Numbers