Databases Reference
In-Depth Information
Table 10.
The final large patterns aster Step 13
L
i
Fuzzy linguistic trends
L
1
(
MH
), (
M
), (
LM
), (
h
), (
mh
), (
m
), (
lm
), (
a
), (
H
)
L
2
(
H, MH
), (
MH,H
), (
h, h
), (
lm, h
), (
H, h
), (
H, mh
), (
H, lm
), (
h, H
), (
H, H
)
L
3
(
MH,h,h
), (
H, h, h
), (
H, lm, h
), (
h, H, h
), (
H, H, h
),
(
h, h, H
), (
H, h, H
), (
h, H, H
), (
h, H, MH
)
L
4
(
h, H, h, H
), (
MH,H,h,H
), (
h, H, h, h
), (
h, H, H, h
), (
H, h, H, H
), (
H, h, H, h
)
L
5
(
H, h, H, H, h
)
600
500
400
300
200
100
0
1.0%
1.5%
2.0%
2.5% 3.0%
Minimum Support
3.5%
4.0%
4.5%
5.0%
Non-removing
Removing
Fig. 3.
The numbers of linguistic trends with and without Step 13
All the linguistic trends found for the class of decreasing trend with sliding-
window size set at 10 and the minimum support set at 0.015 are listed in
Table 10.
In Table 10, the pattern (
H
,
h
,
H
,
H
,
h
) is a linguistic trend with five
fuzzy items in
L
5
. Lowercase and uppercase letters represent the positive and
negative angular degrees upward and downward directions respectively. Most
of the derived linguistic trends in Table 10 have the decreasing property, which
is consistent with the class of
decreasing trend
.
The experiments were then made to compare the numbers of linguistic
trends generated with and without Step 13 of removing redundant large pat-
terns. The results are shown in Fig. 3.
From Fig. 3, it can be easily observed that removing redundant fuzzy large
patterns during the mining process has its e
cacy. Without this step, too
many redundant linguistic trends may be generated and may make users
confused.
