New Algorithm of Maximum Frequent Itemsets Based on FP-Tree for Mining Multiple-Level Association Rules

Discovering maximum frequent item sets is a key problem in data mining. In order to overcome the deficiencies of apriori-like algorithms which adopt candidate itemsets generation-and-test approach, we propose a new algorithm ML_DMFIA which based on DMFIA to mine maximum frequent itemsets in multiple-level association rules. ML_DMFIA utilizes FP-tree structure and up-down progressive deepening searching idea which can avoid making multiple passes over database and does not generate candidate itemsets, consequently, it reduces CPU time and I/O time remarkably. Our performance study shows that ML_DMFIA is more efficient than ML_T2 algorithm for mining both long and short frequent itemsets in mining multiple-level association rules.