Information Technology Reference
In-Depth Information
Table 12 Experiments for sensitivity analysis
S.
No.
Definition
Overall scores (CCi)
i
)
Ranking
A1
A2
A3
1
C1
-
C17
= (1, 1, 3)
0.415
0.402
0.431
A3 > A1 > A2
2
C1
-
C17
= (1, 3, 5)
0.445
0.427
0.459
A3 > A1 > A2
3
C1
-
C17
= (3, 5, 7)
0.456
0.438
0.471
A3 > A1 > A2
4
C1
-
C17
= (5, 7, 9)
0.463
0.444
0.478
A3 > A1 > A2
5
C1
-
C17
= (7, 9, 9)
0.490
0.469
0.506
A3 > A1 > A2
6
C1
= (7, 9, 9), W
C2
-
C17
= (1, 1, 3)
0.429
0.430
0.444
A3 > A2 > A1
7
C2
= (7, 9, 9), W
C1,C3
-
C17
= (1, 1, 3)
0.425
0.400
0.441
A3 > A1 > A2
8
C3
= (7, 9, 9), W
C1
-
C2,C4
-
C17
= (1, 1, 3)
0.427
0.422
0.435
A3 > A1 > A2
9
C4
= (7, 9, 9), W
C1
-
C3,C5
-
C17
= (1, 1, 3)
0.434
0.414
0.441
A3 > A1 > A2
10
W
C5
= (7, 9, 9), W
C1
-
C4,C6
-
C17
= (1, 1, 3)
0.421
0.417
0.441
A3 > A1 > A2
11
W
C6
= (7, 9, 9), W
C1
-
C5,C7
-
C17
= (1, 1, 3)
0.441
0.404
0.441
A3 = A1 > A2
12
W
C7
= (7, 9, 9), W
C1
-
C6,C8
-
C17
= (1, 1, 3)
0.427
0.409
0.444
A3 > A1 > A2
13
W
C8
= (7, 9, 9), W
C1
-
C7,C9
-
C17
= (1, 1, 3)
0.394
0.404
0.432
A3 > A2 > A1
14
W
C9
= (7, 9, 9), W
C1
-
C8,C10
-
17
= (1, 1, 3)
0.429
0.420
0.439
A3 > A1 > A2
15
W
C10
= (7, 9, 9), W
C1
-
C9,C11
-
17
= (1, 1, 3)
0.427
0.380
0.439
A3 > A1 > A2
16
W
C11
= (7, 9, 9), W
C1
-
C10, C12
-
17
= (1, 1, 3)
0.427
0.430
0.448
A3 > A2 > A1
17
W
C12
= (7, 9, 9), W
C1
-
C11,C13
-
17
= (1, 1, 3)
0.405
0.400
0.451
A3 > A1 > A2
18
W
C13
= (7, 9, 9), W
C1
-
C12,C14
-
17
= (1, 1, 3)
0.432
0.413
0.446
A3 > A1 > A2
19
W
C14
= (7, 9, 9), W
C1
-
C13,C15
-
17
= (1, 1, 3)
0.452
0.407
0.431
A1 > A3 > A2
20
W
C15
= (7, 9, 9), W
C1
-
C14,C16
-
17
= (1, 1, 3)
0.425
0.413
0.441
A3 > A1 > A2
21
W
C16
= (7, 9, 9), W
C1
-
C15, C17
= (1, 1, 3)
0.421
0.402
0.437
A3 > A1 > A2
22
W
C17
= (7, 9, 9), W
C1
-
C16
= (1, 1, 3)
0.427
0.430
0.462
A3 > A2 > A1
23
W
C8
= (1, 1, 3), W
C1
-
C7, 9
-
17
= (7, 9, 9)
0.394
0.404
0.432
A3 > A2 > A1
(1/23) and supplier A2 (0/23). Therefore, we can conclude from these results that
our decision process is relatively insensitive to criteria weights with supplier A3
emerging as winner with clear majority. Also, the sensitivities of supplier A1 and
supplier A2 are relatively close as they differ by only one vote.
4.3 Results Validation Using Fuzzy SAW
Tables
13
and
14
present the crisp criteria weights for the 17 criteria and the 3
alternatives. The crisp value (
a) for a fuzzy triangular number
~
a
¼
ð
a
1
;
a
2
;
a
3
Þ
is
obtained using
a
1
þ
4a
2
þ
a
6
.
Table
15
presents the normalized weighted alternative scores for the three
alternatives for the 17 criteria calculated using Eqs. (
15
-
17
). It can be seen that
criteria C8 is a cost type criteria, therefore Eq. (
16
) will be used whereas for the rest
of the criteria Eq. (
15
) will be used for normalization. The overall scores for the
Search WWH ::
Custom Search