Information Technology Reference
In-Depth Information
As elements with these indexes have equal weight coefficients, we obtain
k
+
2
ω
=
,
i
=
1
k
−
1
(4.10)
(
)
i
k
k
+
1
α
<
1
there are several clusters: one cluster of
potency
v
and other clusters of potency 1.
The largest cluster makes a basis, and according to the criterion II the elements
belonging to this cluster are considered to have equal weight coefficients.
According to the criterion I, elements of individual clusters have weight
coefficients less than weight coefficien
ts o
f elements of the largest cluster. Let us
assume that elements with indexes
3.
With confidence level
i
=
1
v
have entered the largest cluster. Let us
rank the elemen
ts whic
h have entered individual clusters by values of indexes
ρ
j
with
ρ
j
=
v
+
1
k
or
j
v
⎡
v
⎤
(
)
(
)
∑
∑
ρ
=
μ
X
,
X
ρ
=
μ
M
,
M
,
⎢
⎣
⎥
⎦
j
R
i
j
j
R
i
j
1
1
i
=
1
i
=
1
or
v
⎡
v
⎤
(
)
(
)
∑
∑
ρ
=
μ
X
,
X
ρ
=
μ
M
,
M
.
(4.11)
⎢
⎣
⎥
⎦
j
R
i
j
j
R
i
j
2
2
i
=
1
i
=
1
of (4.11) are summarized quantity indexes of similarity
(consistency) of elements
ρ
ρ
Indexes
or
j
j
()
i
, entering individual clusters, with
elements from a cluster with potency
v
. Let us consider that the more
X
M
,
i
=
v
+
1
k
i
ρ
(
)
ρ
is,
j
j
(
)
the more importance of element
j
X
is and, accordingly, the more its weight
coefficient is. Let us obtain a conditional ordered series of elements
j
k
Ξ
ranged by
lack of growth of their weight coefficients
X
=
X
=
...
=
X
>
X
>
X
>
...
>
X
(
)
(
)
( )
1
2
v
v
+
1
v
+
2
k
k
or a conditional ordered series of elements
Θ
, ranged in the same manner
M
=
M
=
...
=
M
>
M
>
M
>
...
>
M
.
(
)
(
)
( )
1
2
v
v
+
1
v
+
2
k
(
)
Weight coefficients of elements with indexes
i
=
v
+
1
,
k
are determined by
i
=
v
+
1
k
consecutive substitution of indexes
into the formula (4.7). The result
is
(
)
2
k
−
j
+
1
ω
=
,
j
=
v
+
1
k
.
(4.12)
()
(
)
j
k
k
+
1
