4. EFFICIENT COMPUTATION OF PROB-
k=i P k,j (X ) [Kollios et al]
P j
k=i P k,j−1 (X ) · (1 −
k=i P k−1,j−1 (X) · P(X ⊆ t j )+ P j
ABILISTIC FREQUENT ITEMSETS
P (X ⊆ t j )) [P
≥i,j=0 = ∀.i>j] P(X ⊆ t j ) · P j−1
This section presents our dynamic programming approach,
k=i P k−1,j−1 (X) + ·
(1−P (X ⊆ t j ))· P j−1
which avoids the enumeration of possible worlds in calculat-
k=i P k,j−1 (X )= P (X ⊆ t j )·P ≥i−1,j−1 (X )+
(1 − P (X ⊆ t j )) · P ≥i,j−1 (X ).
ing the frequentness probability and the support distribu-
tion. We also present probabilistic filter and pruning strate-
Using this dynamic programming scheme, we can compute
gies which further improve the run time of our method.
the probability that at least minSup transactions contain
Bạn đang xem 4. - BAI 3 URBAN PLANNING