Information Technology Reference
In-Depth Information
Algorithm-1: Reduct (Minimal number of attributes)
Input:
A decision table
DT
(
C, D
), where
C
: the set of all conditional attributes;
and
D
: the set of decisional attributes;
Processing:
Red←{}
Do
DT
←Red
Loop
(
x C
∈
−
Red)
if
{ }
( )
D
>
( )
D
R x
∪
T
DT
= ∪
Red ←
DT
Until
( ( )
Red {x}
D
=
( ))
D
R
C
Return Red
Output:
Red: A set of minimum attribute subset ; Red ⊆
C
A dataset has at least one reduct in its reduct set—the trivial reduct (i.e., the dataset itself). It also
has one or more minimal reducts.
r ule generation and Building the Classifier
The main task of the rule generation method is to compute reducts relative to a particular kind of infor-
mation system. The process by which the maximum number of condition attribute values is removed
without losing essential information is called value reduction, and the resulting rule is called maxi-
mally general or minimal length. Computing maximally general rules is of particular importance in
knowledge discovery since they represent general patterns existing in the data. In this subsection, we
discuss a method to simplify the generated decision rules by dropping some condition attributes. The
rule generation algorithm is described as follows: (Refer to Algorithm-2).
Algorithm-2: Rule Generation Algorithm
Input:
A set of specific decision rules RULE
Processing:
GRULE ←Φ
N ←|RULE|
For i=0 to N-1 do
r
←
r
i
M
← |
r|
For j = 0 to M-1 do
Remove the j
th
condition attribute a
j
in rule
r
If
r
inconsistent with any rule
r
∈
RULE then restore the dropped condition a
j
End if
End for
Remove any rule
r
∈
GRULE that is logically included in rule
r
If rule
r
is not logically included in a rule
r
∈
GRULE then
GRULE ←
r
GRULE
Search WWH ::
Custom Search