基于FP_tree的最大频繁模式挖掘算法的改进

Improved Algorithm for Mining Maximal Frequent Patterns Based on FP_tree

基于条件模式树的最大频繁模式挖掘算法在挖掘过程中将扫描事务数据库两次,且产生了大量的候选项目集,产生最大频繁模式过程中比较次数较多,总体效率较低.提出改进后的最大频繁模式挖掘策略,利用二维表保存事务出现项目的情况,通过最大频繁模式的相关性质减少了挖掘的项数及产生的频繁模式集,减少比较的次数.

The maximal frequent pattern mining algorithm based on the conditional pattern trees scans the transaction database twice and produces a large number of candidate itemsets in the mining process. During the process of generating the maximal frequent patterns, there are more comparisons and lower overall efficiency. Using two-dimensional table to save the items which appear in the transaction, by means of the relevant properties of maximal frequent patterns, this paper reduces the items mined and the frequent patterns generated. The times of comparison decreases. The improved maximal frequent pattern mining strategy is presented.