Copying or distributing in print or electronic forms without written permission
of Idea Group Inc. is prohibited.
association rule (CMAR) r drawn from C is an implication of the form
Y1 ?‡’ Y2. The support of rule r is:
1 2
1
( ) ???
=
= ??‘
??‘ ??• j f x j
n j
m
supp Y Y
mj ,
where n is the total number of facts, and mj is a count value associated
with fact fj in X. The confidence of r is computed as the ratio supp(Y1???Y2)/
supp(Y1), where:
1
1
1
( ) ???
=
= ??‘
??‘
j f Y j
n j
m
supp Y
mj for all facts fj supporting Y1.
Definition 2: A NoN-couNt-bAseD multiDimeNsioNAl association rule
(NCMAR) r generated from a cube C =
is an implication of the
form Y1 ?‡’ Y2, where Y1??? Y2 is a meaningful FCI and Y1??© Y2 = ??…, and
??? m a non-count measure in M such that m ??? Y1 ???Y2. Let the set of facts
supporting Y1 ???Y2 be X. Then, the support of rule r is:
1 2
1
( ) ???
=
= ??‘
??‘ ??• j f x j
n j
m
supp Y Y
mj ,
where n is the total number of facts, and mj is a count value associated
with fact fj in X. The confidence is computed as the ratio supp(Y1???Y2)/
supp(Y1), where:
1
1
1
( ) ???
=
= ??‘
??‘
j f Y j
n j
m
supp Y
mj for all facts fj supporting Y1.
It is important to note that in the case of NCMARs, the cube contains a non-countbased
measure apart from a count-based one. The latter is used to compute the
support of the rule.
Pages:
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492