HANA algorthim : a novel algorithm for frequent itemsets generation

Abstract EN

This paper proposes a novel algorithm, HANA algorithm, to generate frequent item sets and their subsequent in preparation of data mining.

The concept behind the algorithm is that the common item sets between / among transactions are the candidate for frequent item sets with a known support that is equal to the number of involved transactions.

The proposed algorithm consists of two processes; the first one is to extract candidate frequent item sets (K-item sets) and the second is to generate their subsequent (k-item sets).

HANA algorithm in this context is used to gain more insight and to study issues related to the mining process.

All item sets and all their subsequent with all possible support values are extracted.

The experiment is conducted on three synthesized binary datasets with different density and several interesting issues are raised.

HANA algorithm has several advantages among them is that it reads the dataset only once, there is no need to count the support of k-item sets, it reduces the computation needed by the choice of the matrix representation and removing duplicates of item sets and finally, K-item sets subsequent generation is independent on the K items.