摘要
针对Apriori算法的不足,提出一种基于粗糙集的频繁项集发现方法。新方法首先利用粗糙集的特征属性约简算法进行属性约简,找到其中的核心属性数据,然后在构建约简决策表的基础上应用改进的Apriori算法对这些核心数据进行数据挖掘,最终得到频繁项集。改进方法的优势在于在保证知识库分类能力不变的前提下消除不必要的冗余属性,减少了属性数目。在生成频繁项目集方面,根据k-1频繁项集中的项目个数来确定是否生成对应的k-候选集,这样就不需要进行连接操作生成k-候选集,减少了候选项集的生成数量。实验验证了所提出的改进算法的有效性,尤其在k很大的时候,可以节省大量的计算时间,避免产生大量的候选集,可显著提高数据挖掘的效率。
Aiming at the defects of the Apriori algorithm, this paper proposed a method for the discovery of fre- quent item set based on rough set. Firstly, this method uses characteristics attributes reduction algorithm of rough set to find the core of the attribute data, and then uses the improved Apriori algorithm on the data mining of these core data based on the reduction decision table to get frequent item set. The advantage of this method is that it can elimi- nate unnecessary attributes and reduce the number of attributes on the premise of the same knowledge base classifica- tion ability. In terms of generating frequent item set, whether the k-candidate set is generated is determined by the number of k-1 frequent item set, In this way the number of candidate item set can be reduced. The experiment veri-fies the validity of the improved algorithm, especially when k is a great number. It can save a lot of computing time, avoid a lot of candidate set, and improve the efficiency of data mining significantly.
出处
《计算机仿真》
CSCD
北大核心
2013年第1期329-332,385,共5页
Computer Simulation
关键词
粗糙集
候选集
频繁项集
Rough set
Candidate set
Frequent item set
