AMKIS: An Algorithm for Association Mining

Mining frequent items and itemsets is a daunting task in large databases and has attracted research attention in recent years. Generating specific itemset, K –itemset having K items, is an interesting research problem in data mining and knowledge discovery. In this paper, we propose an algorithm for finding K itemset frequent pattern generation in large databases which is named as AMKIS. AMKIS algorithm uses no candidate generation and minimum support criteria’s for generating K itemset frequent pattern. The structure and functionality of AMKIS is different from Apriori and FP tree based algorithms. Further, this algorithm does scan transaction database once. AMKIS performance is compared with Apriori algorithm. Our extensive performance study shows that AMKIS algorithm has higher performance as compared with Apriori. The proposed algorithm, AMKIS, is highly scalable for mining not only small but also large K itemset frequent patterns and is linearly scalable in terms of the database size. Keywords— association mining, association rules; business intelligence; data mining; frequent itemset; frequent patterns; knowledge discovery; k itemset and mining methods.