一种基于FP-Growth的改进算法

An improved algorithm based on FP-growth

关联规则挖掘由于表达形式简洁、易于解释和理解已成为数据挖掘中的研究热点,对关联规则的研究具有重要的理论价值和现实意义。文章分析频繁项集挖掘算法FP-growth算法,针对算法中存在的效率瓶颈问题,提出了一个改进的挖掘算法。改进后的算法通过投影统计的方法直接得到频繁1-项集的条件模式基,从而减少了FP-growth算法中构造FP-tree和搜索的开销。通过分析,说明改进的算法具有良好的性能。

Association rule mining has become an important and active research topic in the data mining due to its succinct, easy to be explained and understood. It has important theoretical value and the practical significance to study the association rules. After analyzing FP- growth association mining algorithm, this paper proposes an improved algorithm to overcome the low efficiency of FP- growth algorithm by adopting the projection method. The improved algorithm can directly obtain the condition pattern of frequent 1 - item, thus it reduces the time cost of building the FP- tree in the FP- growth algorithm. It proved that the improvement algorithm has higher performance than orginal FP- tree algorithm.