Civil Engineering Reference
In-Depth Information
The weight of factors for an object is different from person to person, due to the
subjectivity of personal preference. In a diverse society, knowledge accumulated
day and day, the change of environment, volatility of information and complicated,
diverse, fuzzy and indeterminate human behavior. Therefore, how to decide the
weights, called fuzzy weights, becomes a primary work before evaluating the
specialized human capital. In this section, we will demonstrate an integrated design
via appropriate questionnaires of field study to reach a common agreement for
weight of fuzzy factors for an object/event.
2.3
Defuzzification with Fuzzy Data
Once such a transformation has been selected, instead of the original trapezoid
data, we have a new value
y
. In an ideal situation, this new quantity
y
is
normally distributed. (In practice, a normal distribution for
y
may be a good first
approximation.) When selecting the transformation, we must take into account that,
due to the possibility of a rescaling, the numerical values of the quantity
x
is not
uniquely determined.
=
f
(
x
)
Definition 2.1.
Defuzzification for discrete fuzzy data. Let
X
be a fuzzy sample on
universe domain U with ordered linguistic variable
L
i
:
i
=
1
,
2
,...,
k
corresponding
i
to integral values,
μ
x
(
L
i
)=
m
i
be the membership with respect to
L
i
,
Σ
μ
x
(
L
i
)=
1.
=
1
1
k
−
1
Σ
i
j
Let
c
x
=
Σ
1
m
i
L
i
be the centroid of the fuzzy data, and
dx
=
1
m
i
|
i
−
cx
|
be
=
=
its deviation for the
cx
. We call
Xf
=
cx
+
dx
, the defuzzification value for the
discrete fuzzy sample
X
.
0
1
+
0
.
6
2
+
0
3
3
+
.
0
1
4
+
.
0
5
Example 2.1.
Let
X
=
be a discrete fuzzy sample on the
ordered linguistic variable
{
L
1
=
1
,
L
2
=
2
,
L
3
=
3
,
L
4
=
4
,
L
5
=
5
}
. Then the
defuzzification value for the fuzzy data
X
is
1
i
j
Xf
=
Σ
1
m
i
L
i
+
1
Σ
1
m
i
|
i
−
cx
|
=
2
.
5
+
0
.
15
=
2
.
65
=
=
k
−
3
An Integrated Fuzzy Evaluation Process
On the above-mentioned human capital measurement, we will consider a new
approach of measurement. This is because analysis by traditional methods usually
involves the following weaknesses: (a) the use of arithmetic in traditional question-
naires is often overexplained. (b) Experimental data is often overused just to cater
to the need for apparent numerical accuracy. (c) For the sake of simplifying the
evolutional model, the relationship of actual condition and dynamic characteristic is
neglected. We have to better make use of fuzzy statistical technique at investigation
realm to estimate the human resource capital.
Search WWH ::
Custom Search