Mining Frequent Patterns based on Compressed FP-tree without Conditional FP-tree Generation
Fei Chen, Lin Shang, Ming Li, Zhao-Qian Chen, Shifu Chen · 2006
Frequent patterns mining are widely used in many practical data mining applications. Therefore, current research focuses on developing frequent patterns mining algorithms of high performances and FP-growth is proved as an important and efficient frequent patterns mining algorithm. In this paper, a new algorithm Temporary Root growth based on Compressed FP-tree, i.e. TR-CFP, is proposed. This algorithm employs a temporary root constructing thought during mining on a CFP- tree without conditional FP-tree generation. TR-CFP saves large memory space occupied by FP-tree and the cost of constructing many conditional FP-trees. Experiments show that the time and space for TR-CFP have reduced significantly compared to FP- growth mining based on FP-tree. Furthermore, TR-CFP has a particular character other than Apriori and FP-growth, that it can specially mine frequent patterns of the designated length dynamically and efficiently. In further experiments, even at a low support threshold, when the whole mining process takes a long time, special mining of TR-CFP is still very quick. This particular character may be very useful in applications.