When asked to "find all frequent Item sets" is it just the set that was last worked out that is the answer you should give or is it necessary to give all before that too.
For example, say if the last results I get for a set is (A,B,D) then is that my frequent item set or do I need to also include all found before that that also satisfy minSup, e.g., (A,B) (A,D) (B,D), etc?
Frequent itemset mining as performed by the original Apriori algorithm generates all frequent itemsets. The Downward Closure Property says that:
Every non-empty subset of a frequent itemset is also frequent.
So if {(A,B,D)} is frequent then so are {A}, {B}, {D}, {(A,B)}, {(A,D)} and {(B,D)}. Apriori algorithm will find all these frequent itemsets.
You do want to know all frequent itemsets, because you may need them for association rules.
I.e. even if your maximal frequent itemset is A, B, C, D you do want to remember the support of A,B,C; A,C,D; A,B; and so on, because you need this information for computing confidence.
Look up the technique called "association rules," and the a priori algorithm. This is also known as "market basket analysis." Frequent item sets gets complex and depends on the the data.
