Pattern Discovery - Association Rules

confidence: the conditional probability of pattern Y appearing when pattern X appears. The formula for confidence is the support of the union of patterns X and Y, divided by the support of pattern X:

s(X ∪ Y) / s(X)

minconf: minimum confidence, below which a potential association is uninteresting or meaningless.

The association rule is

X → Y (s, c)

That is, X implies Y when support(X ∪ Y) >= minsup and confidence(X, Y) >= minconf.

Both X and Y must be k-itemsets where k > 1.

minsup: probability of itemsets
minconf: probability of an association rule between itemsets