Databases Reference
In-Depth Information
Tabl e 5 .
The set of large 1-pattern-sets
L
1
for this example
Itemset
Count
Itemset
Count
Itemset
Count
P
1
.h
2.26
P
3
.M
2.00
P
5
.h
2.04
P
1
.M
2.00
P
3
.mh
0.96
P
5
.M
1.00
P
1
.m
1.00
P
4
.M
2.00
P
6
.H
2.04
P
2
.h
2.96
P
4
.H
1.23
P
6
.h
2.04
P
2
.M
2.00
P
4
.h
2.04
P
6
.M
1.00
P
3
.h
2.71
P
5
.H
1.93
Tabl e 6 .
The membership values for
P
1
.H ∩ P
2
.H
s
p
P
1
.h
P
2
.h
P
1
.h ∩ P
2
.h
1
0
0
.
92
0
.
0
2
0
.
92
0
.
67
0
.
67
3
0
.
67
0
0
.
0
4
0
0
0
.
0
5
0
0
0
.
0
6
0
0
.
67
0
.
0
7
0
.
67
0
.
70
0
.
67
2-pattern-sets are generated. Note that no two fuzzy terms with the same
P
i
are put in a candidate 2-pattern-set.
STEP 10: The following substeps are done for each newly formed candidate
pattern-set.
STEP 10.1: The fuzzy membership value of each candidate pattern-set in
each subsequence is calculated. Here, assume the minimum operator is used
for the intersection. Take (
P
1
.h
,
P
2
.h
) as an example. The derived membership
value for this candidate 2-pattern-set in
s
2
is calculated as: min(0.92, 0.67) =
0.58. The results for the other subsequences are shown in Table 6.
STEP 10.2: The scalar cardinality (count) of each candidate 2-pattern-set
in the subsequences is then calculated.
STEP 10.3: The supports of the above candidate pattern-sets are then
calculated and compared with the predefined minimum support 0.075. In
this example, 18 pattern-sets satisfy this condition. They are thus kept in
L
2
(Table 7).
STEP 11: Since
L
2
is not null in the example,
r
=
r
+ 1 = 2. Steps 8-10
are then repeated to find
L
3
and others. In this example, the other fuzzy large
pattern-sets found are shown in Table 8.
STEP 12: The large patterns are shifted to the ones with the first data-
point subscript. For example, the three patterns (
P
2
.h
,
P
3
.h
,
P
4
.h
), (
P
3
.h
,
P
4
.h
,
P
5
.h
)and(
P
4
.h
,
P
5
.h
,
P
6
.h
)in
L
3
are shifted into (
P
1
.h
,
P
2
.h
,
P
3
.h
).
The other patterns are also checked for shifting in the same way.
STEP 13: Redundant patterns are removed. For example, the four large
patterns, (
P
1
.h
,
P
2
.h
,
P
3
.h
), (
P
1
.h
,
P
2
.h
,
P
3
.h
), (
P
1
.h
,
P
2
.h
,
P
3
.h
)and(
P
1
.h
,
P
2
.h
,
P
3
.h
), are the same and only one of them is kept. The final results are
shown in Table 9.
