Efficient Improvement of FT-Tree Based Frequent Itemsets Mining Algorithms

The I/O cost of database scanning has been a bottleneck problem in data mining. Many algorithms proposed recently are based on FP-tree. These algorithms include all frequent itemsets mining, closed frequent itemsets mining and top-k closed frequent itemsets mining. However, creating FP-tree from database must scan database two times. In order to enhance the efficiency of FP-tree based algorithms, using buffer and merging methods, a novel algorithm called QFPC was proposed which can create FP-tree with one database scanning