I'm trying to find the optimal way of identifying common patterns with groups of items (an item is identified by a number)
e.g. every transaction has a list (of size 1 - 20) of items. I can easily say which single items appear most often by keeping a count and incrementing per transaction. I guess i could use the same approach for combinations of items (2+) that appear in transactions, but i'm not sure if this is the best approach?
note that order of the item list is not important, but that i'm not looking to treat the full list as a unique transaction (I want to split each list into its sub combinations - ab,ac,bc, abc, etc and match on these)
any help appreciated,