Databases Reference
In-Depth Information
Tabl e 3 .
Time needed to extract rules and action rules by
DEAR
DataSet
Rules
Action rules DEAR
Breast Cancer
20
s
27
min
51
s
Cleveland
1
min
09
s
Over
8
h
Hepatitis
54
s
Over
8
h
a
b
c
{x
1
,x
2
,x
3
,x
4
}
0
{x
2
,x
4
}
R
{x
1
,x
3
}
0
{x
2
,x
4
}
1
{x
5
,x
6
}
P
Table 3
c = ?
c = 1
c = 0
a
b
a
b
a
b
{x
1
,x
2
,x
3
,x
4
}
{x
1
,x
3
}
0
{x
2
,x
4
}
{x
2
,x
4
}
R
P
Table T2
Table T3
{x
5
,x
6
}
Table T1
a = 0
a = 0
b
b
{x
1
,x
2
,x
3
,x
4
}
{x
2
,x
4
}
{x
5
,x
6
}
R
P
Table T4
Table T5
Fig. 1.
(
d, L
)-tree
a
b
c
{x
7
,x
8
}
{x
7
,x
8
}
2S
S2
c = 2
Table 4
c = ?
a
b
a
b
{x
7
,x
8
}
{x
7
,x
8
}
S
S
Table T7
Table 6
Fig. 2.
(
d, H
)-tree
Now, it can be checked that only pairs of rules belonging to tables
{
can be used in action rules
construction. For each pair of tables, we use the same algorithm as in [11] to
construct extended action rules.
This new algorithm (called
DEAR-2
) was implemented and tested on many
datasets using PC with 1.8 GHz CPU. The time complexity of this algorithm
was significantly lower than the time complexity of the algorithm
DEAR
pre-
sented in [11]. Both algorithms extract rules describing values of the decision
[
T
5
,T
7]
,
[
T
5
,T
6]
,
[
T
2
,T
6]
,
[
T
3
,T
6]
,
[
T
4
,T
7]
}
