Font Size: a A A

The Algorithm Of Maximum Mining Frequent Super Sets Based On The FP-tree

Posted on:2012-03-06Degree:MasterType:Thesis
Country:ChinaCandidate:J WangFull Text:PDF
GTID:2218330335976298Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Data mining (DM), arising in the 1980s, is constantly studied in order to solve the problem that data volume is so enormous without good quality invariably.There are some potential links between the data. In the database it exist a class of important knowledge that can be found.1993, Agrawal proposed mining association rules between items in the database. Since then, many researchers did a lot of research about mining association rules, and made a lot of the classic algorithm.Many of these algorithms have their drawback in the further research, which can produce so many candidate sets after scanning the data base repeatedly that there will be the waste of both time and memory space. To solve the drawback, Han with his research fellows presents a FP-growth algorithm. This thesis gives a new idea under the foundation of FP-growth algorithm.1.Sturcture IFP-tree. Scan the database, the frequent itemsets compressed into a FP-tree, which retain the association between itemsets. Each transaction in the frequent item sets insert into the process of FP-tree, the dynamic pointer to achieve, which can improve storage utilization.2. Improved maximum frequent pattern tree (MMFIT). In MFIT, the path which starting from the root to a leaf node stands for a global frequent itemsets. The middle node is the length of root node to the path. In the super set checking, you can quickly interview the head table that contains the itemsets which corresponding to the path of frequent itemsets, and then tested the project from the bottom up and turn the set of matching items.This algorithm does not need produce numerous candidate sets, at the mean time reduces the frequency of scan data set and the database traversal, which improves efficiency of the algorithm. Experiments results show that the algorithm in reducing redundant candidate itemsets while effectively reduce the algorithm running time.With the development of information technology, the application of DM will be broader with providing helps and guiding significance in banking, sales and other commercial activities. The service from DM should be the key point in the future work.
Keywords/Search Tags:Association Rules, Frequent Pattern, FP-growth, FP-tree, IFP-tree
PDF Full Text Request
Related items