An Efficient Algorithm for Mining Maximal Frequent Sequential Patterns in Large Databases

Frequent sequence mining is one of the important research directions of click stream analysis, this paper studies the problem of mining maximal frequent sequences in mobile app clickstreams. Different from frequent itemsets mining, frequent sequence mining takes the time order of elements into account. In this paper, MFSGrowth(Maximal Frequent Sequence Growth) is proposed for fast discovery of frequent sequence. MFSGrowth is an efficient algorithm based on the FP Tree, combined with the storage characteristics of the TriedTree, experiments show that the algorithm performs well in both mining time and storage efficiency.