基于fp树关联规则算法改进的研究

《基于fp树关联规则算法改进的研究》由会员上传分享，免费在线阅读，更多相关内容在学术论文-天天文库。

1、摘要数据挖掘是从海量数据中发现蕴含其中的有用信息和知识的过程，应用广泛。关联规则分析作为数据挖掘的主要功能之一得到了广泛应用，对关联规则算法的深入研究具有很高的理论价值和现实意义。本文首先阐述了数据挖掘的相关知识，对关联规则经典算法Apriori算法和FP-growth算法作出详细介绍，分析并指出了这些经典算法存在的问题与不足。虽然基于FP-Tree的经典FP-growth算法不需要像Apriori算法一样产生大量的候选频繁项集，但必须进行大量的遍历和递归操作，严重影响了挖掘的整体效率。针对FP-growth算法在构建FP-Tree和挖掘频繁项集

2、两方面存在的缺陷，本文提出了DFP-Tree构造和FP-数组技术来提高算法的效率，结合两种技术提出DFPmine算法，并通过实验证明了算法的效率。实验结果表明，DFPmine算法比FP-growth算法来的更有效率，尤其是在数据集比较大且比较稀疏，关联聚集度较低，FP-Tree分支较多的情况下，本算法效果明显，比经典FP-growth算法能获得更好的性能。关键字：关联规则;FP-Tree;FP-growthIAbstractDataminingistheprocessoffindingtheusefulinformationandknowledg

3、einthehugeamountsofdataforpeople,appliedwidely.Associationrulesasastrongfunctionofdatamining,researchonwhichhasahightheoreticalvalueandpraticalsignificance.Thisarticlefirstlydescribesthedataminingknowledgebriefly,givesdetaileddescriptionoftheclassicassociationrulesalgorithmss

4、uchasApriorialgorithmandFP-growthalgorithm,illustratesandpointsouttheshortcomings.AlthoughtheclassicFPalgorithm,basedontheFP-Tree,doesnotneedtocreatealargenumberofcandidatefrequentitemsetsastheApriorialgorithm,itneedsalargenumberoftraversalandrecursionoperations,whichreducest

5、heefficiencygreatly.ToovercomethedefectsoftheFP-growthalgorithminbuildingtheFP-Treeandminingfrequentitemsets,anewstructureofDFP-TreeandFP-arraytechnologyareproposedtoimprovetheefficiencyofthealgorithm.Basedonthem,DFPminealgorithmisproposed.Then,doexperimentstoprovetheefficien

6、cyofthealgorithm.TheexperimentsshowthattheDFPminealgorithmismoreefficientthantheFP-growthalgorithm,inparticular,itismoreeffectiveandhasbetterperformancethantheclassicFP-growthalgorithmwhenthedatasetsarerelativelylargeandsparse,theassociationaggregationdegreeislow,ortheFP-Tree

7、hasagreatnumberofbranches.Keywords：associationrules;FP-Tree;FP-growthII目录摘要..............................................................................................................................IAbstract..................................................................

8、......................................................II目录..........