An Efficient Association Rule Mining Using the H-BIT Array Hashing Algorithm

Association Rule Mining (ARM) finds the interesting relationship between presences of various items in a given database. Apriori is the traditional algorithm for learning association rules. However, it is affected by number of database scan and higher generation of candidate itemsets. Each level of candidate itemsets requires separate memory locations. Hash Based Frequent Itemsets Quadratic Probing (HBFI QP) algorithm, which is based on hashing technique for mining the frequent itemsets. In order to stay away from collisions and primary clustering in hashing process, Quadratic Probing (QP) technique is used. Though the primary clustering and collisions are eliminated, secondary clustering is formed in all cases and the hash table occupies more space than the total number of items in the database. To avoid those problems, the H-Bit Array Hashing (H-BAH) algorithm is presented in this paper. HBAH algorithm reduces hash table size required for placing items and it also avoids hash collisions and secondary clustering. The H-Bit array that is added to the first or header bucket of the table gives the information about which buckets are hashed initially. At the time of collisions in the hashing process, the H-BAH algorithm works by finding the neighbourhood of buckets near the original hashed bucket, in order to place the collided items quickly. The HBAH algorithm provides frequent itemsets with less computational time and memory than the existing algorithm.