基于FP-tree最大频繁模式超集挖掘算法

Maximal Frequent Pattern Superset Mining Algorithm Based on FP-tree

数据挖掘应用中的最大频繁项集挖掘算法大多存在候选项目集冗余问题,造成时间和空间的浪费.针对此问题,通过构造条件FP-tree,对不符合要求的项目进行剪除并对MFIT算法进行改进,提出一种基于FP-tree的最大频繁模式超集挖掘算法.此算法无需产生大量的候选集,同时减少数据集扫描次数,降低数据库遍历时间,提高算法效率.实验证明,此算法在降低候选项目集冗余度的同时有效减少了算法运行时间.

The main problem existing in maximal frequent itemsets mining algorithms of data mining applications was candidate set redundancy,waste of time and space.The constructed conditioning FP-tree would cut off the items which did not meet the requirements and improve MFIT algorithm.The conditioning FP-tree was proposed as the largest frequent pattern superset mining algorithm based on FP-tree.This algorithm did not produce numerous candidate sets,at the mean time,reduced the frequency of scan data set and the database traversal,improving efficiency of the algorithm.Experiments results showed that the algorithm reduced redundant candidate itemsets and effectively decreased the algorithm running time.