Question:
Consider an itemset I = X ∪Y = X′ ∪Y′ and let X′ ⊆ X. Show that confidence of the association rule
X ⇒ Y is greater than or equal to that of X′ ⇒ Y ′.
Answer:
Given: I=X∪Y=X′∪Y′ and X′⊆X
Proof: Since, I=X∪Y=X′∪Y′, we know: Support(X∪Y) = Support(X′∪Y′)
This is because represent the same set of items, so their support must be equal.
Given , it follows that: Support(X′) ≥ Support(X).
This is because is a subset of , meaning every transaction that contains will also contain .
The number of transactions containing X' is at least equal to those containing X.
Comparing Confidence:
- Confidence of : Confidence(X⇒Y) = Support(X∪Y) / Support(X)
- Confidence of : Confidence(X′⇒Y′) = Support(X′∪Y′) / Support(X′)
Since, and we have:
Confidence(X⇒Y)= Support(X∪Y)/Support(X) ≥ Support(X∪Y)/Support(X′) = Confidence(X′⇒Y′)
The confidence of the association rule when .