基于偏微分方程的关联特征数据挖掘改进方法研究  

Research on PDE⁃based improved method for correlation feature data mining

作　　者：王新龙[1] WANG Xinlong(Changzhi University,Changzhi 046011,China)

机构地区：[1]长治学院,山西长治046011

出　　处：《现代电子技术》2021年第18期111-113,共3页Modern Electronics Technique

基　　金：国家自然科学基金项目(61773002)。

摘　　要：在关联特征数据挖掘的过程中,常因数据集规模过大,使空间复杂度过高,导致数据资源紧缺,为此,文中提出基于偏微分方程的关联特征数据挖掘改进方法研究。构建多叉树,根据其内部属性,划分等价类数目,生成频繁项集,读入原始事务集,改良负载均衡,实现关联特征数据挖掘改进方法研究。设计对比实验,选用多组数据集测试两种方法的空间资源占用情况,实验结果表明,改进后的数据挖掘方法在空间复杂度上的增长始终保持稳定,能够满足数据挖掘需求,达到了预期的研究目的。In the process of correlation feature data mining,because of the extremely⁃large scale of data set,the spatial complexity is too high,which leads to the shortage of data resources.Therefore,a PDE⁃based improved method of correlation feature data mining is proposed for the research.According to its internal attributes,the number of equivalent classes is divided,the frequent item sets are generated,the original transaction set is read into it,the load balancing is improved,and the improved method of correlation feature data mining is realized.A comparative experiment is designed to test the occupancy of space resources of the two methods on multiple data sets.The experimental results show that the improved data mining method keeps stable in growth of space complexity and can meet the needs of data mining.This method achieves the expected research purpose.

关 键 词：关联特征 数据挖掘 改进方法 偏微分方程 多叉树构建 负载均衡 

分 类 号：TN919-34[电子电信—通信与信息系统] TP311.13[电子电信—信息与通信工程]