Practical Applications of Data Mining

2.3 attriBute-orienteD rule generalization 43

while (# of distinct values in A

i

) > threshold do {

if there is no higher-level concept for A

i

in the

concept hierarchy table

then remove A

i

else

(1) substitute the values of A

i

with its

corresponding minimally generalized

concept;

(2) merge the identical tuples;

}

while (# of tuples in GR) > threshold do {

(1) selectively generalize attributes;

(2) merge the identical tuples;

}

}

end {the end of basic attribute-oriented induction}

Step 3: Simplify the generalized relation.

Step 4: Transform the final relation into a logical rule.

The basic attribute-oriented induction algorithm ...