An Improved FP-Tree Algorithm for Mining Maximal Frequent Patterns

FP-growth algorithm produces all the frequent item sets without producing a large number of candidate items. However, when the item set is too large, the branch of the spanning tree will be long, occupying the space too large, and the mining efficiency will be reduced. In this paper, the method of using dynamic insert node FP - tree structure, and all the back pointer, to generate a new type of FP - tree. This article also proposes Max-IFP maximum frequent patterns mining algorithm, using the new generation of FP - tree dug up all the maximum frequent item sets. The experimental results show that the new FP-tree occupies a smaller space, and the algorithm proposed in this paper is shorter and more effective than other algorithms when mining the maximum frequent item sets.