基于改进的FP-树和数组技术的频繁模式挖掘算法  

A Improved Algorithm of Frequent Patterns Mining Based on FP-Tree and Array

作　　者：孟志忠[1] 

机构地区：[1]太原科技大学计算机科学与技术学院,太原030024

出　　处：《电脑开发与应用》2012年第9期43-45,51,共4页Computer Development & Applications

摘　　要：FP-growth算法是目前较高效的频繁模式挖掘算法之一,该算法不产生候选项集,但递归构造"条件FP-Tree"的CPU开销和存储很大。为此提出了一种频繁模式挖掘算法IFPmine。首先,为了节省内存空间,采用了约束子树的挖掘方法;其次,采用了数组技术来减少树的遍历时间,从而提高算法的效率。实验结果表明,IFP算法是一种较有效的频繁模式挖掘算法,其挖掘效率优于树算法和树算法而需要的内存却少于树和树算法。FP-growth algorithm is an efficient algorithm for mining frequent patterns, which doesn＇t generate candidate item set,So the algorithm has higher efficiency, But recursive structure conditions FP-Tree CPU cost and storage is costly. For all this this paper presents a frequent pattern mining algorithm. Fristly using constrained subtrees of a compact IFP-tree to mine frequent pattern is used, reduces the memory consume; Secondly, an array-based technique to reduce the traverse time to the IFP-tree is used. The experimental evaluation shows that IFP algorithm is a high performance algorithm. The performance is better than STFP-tree algorithm and FP-tree algorithm, and need memory less than STFP-tree algorithm and FP-tree algorithm.

关 键 词：数据挖掘 关联规则 频繁模式 IFP-树 

分 类 号：TP311[自动化与计算机技术—计算机软件与理论]