An efficient maximal frequent itemset mining algorithm based on linear prefix tree

摘要

Among various problems in data mining, discovering maximal frequent item-sets is the most crucial one. Mining all the frequent itemsets will generate big amount of item-sets. Frequent maximal itemsets (FMIs) results in a much smaller number of itemsets. Hence, it is highly valuable to explore maximal frequent itemsets. In general, exploration of frequent itemsets has been implemented using a special data structure called LP-tree (Linear Prefix-tree). The presentation of LP-tree is in array form which reduces the usage of pointers among nodes. Linear prefix tree uses less information and linearly accesses corresponding nodes. In this research paper, a novel technique is designed called LP-MFI-tree for mining maximal frequent itemsets (MFI) which is extended from LP-growth method. It reduces memory consumption and runtime. Then, the performances of the LP-MFI-tree are validated through various experiments on different datasets.