Mining frequent pattern based on fading factor in data streams
Jiadong Ren, Hui-Ling He, Changzhen Hu, Lina Xu, Libo Wang · 2009
In order to improve the mining efficiency of frequent patterns in data streams, we present an algorithm DS-FPM for mining frequent patterns in data streams. First, a data structure DSFP-tree is constructed and the data stream is divided into a set of segments, then potential frequent itemsets on each segment are obtained by IGFA algorithm, while the generated itemsets and the remaining itemsets of DSFP-tree generated by the earlier segment and sampled by fading factor are stored in new DSFP-tree, finally, the frequent patterns in the data stream can be rapidly found by a breadth-first search strategy. The experimental result shows that the execution efficiency of DS-FPM is better than that of FPIL-STREAM algorithm.