Practical Applications of Data Mining

2.6 Mining of CoMpaCt rules 63

2.6.4 Learning Algorithm

In this section, the steps to derive rules using SADs are described with an ex-

ample. First, it is assumed that there are k condition attributes (CA

1

, CA

2

, …,

CA

k

) and a Target Attribute (TA) in the generalized relation. The steps to

derive the rules are as follows:

Step 1 (First iteration):

Compute the SAD from each condition event to the target event.

Step 2 (Iterations 2 through k):

For each subsequent iteration i (2 ≤ i ≤ k), compute the SAD from each

combination of j − i + 1 condition events CE = {(CA

i

, ca

i

), (CA

i+1

, ca

i+1

), …,

(CA

j

, ca

j

)} to each target event with the following ...