欢迎来到天天文库
浏览记录
ID:36353379
大小:4.78 MB
页数:121页
时间:2019-05-09
《基于粗糙集的数据挖掘与决策支持方法研究》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、浙江大学硕士学位论文基于粗糙集的数据挖掘与决策支持方法研究姓名:苏健申请学位级别:硕士专业:计算机应用技术指导教师:高济2002.12.1条件向量),并得到条件向量的约简。经过约简以后的条件向量能够为决策者提供更好的选择。粗糙决策支持方法能够充分挖掘决策表本身潜在的决策支持能力,以提供强有力的决策支持,并在决策支持过程中强调决策者的参与,具有棼大的灵活性。粗糙决策方法能够与传统方法组合成优势互补的混合决策支持模式。2户。关键词:粗糙集理论j数据挖掘i决策支持j属性约简歹规则获黔AbstractRecently,datamillinghasbeenoneofhotrese
2、archarea.Dataminingaimstoextractimplicit,previouslyunknownandpotentiallyusefulknowledgefromlargeamountofdata.Theknowledgediscoveredbydata
3、癫砖珏gtechnologiescanbeused协offerdecisionsupport。Roughsettheoryisanimportantmathematicaltoolforuseincircumstanceswhicharecharacterizedbyvaguenessandunce
4、rtainty.IthasbeenproventobeveryusefuIinthefieldofdatamining.Theresearchworkinthisdissertationisbasedonthefollowingobservations:(1)mostexistingroughsetmethodslackofsuitablemeanstodealwithdistribmeddataenvironment;(2)sincethedecisionsupportabilityofdecisiontablewillbereducedinruleinduction
5、process,theobtainedrulesCallonlyofferlimiteddecisionsupportcompared、析ththedecisiontable.Inordertoresolvethesetwoproblems,meta-informationmethodandroughdecisionsupportmethodareproposedrespectively.Meta。informationmethodaimstoprovideacost-efficientmeanstodealwi伍distributeddataenvironment.M
6、eta-informationmethodconsistsofmeta—informationgenerationandmaintenancemethodandroughsetmethodbasedonmeta-informafion。Inmeta-informationmethod,theconceptofinformationsysteminoriginalroughsettheoryisgeneralizedbyconsideringafamilyofsubsystems。Thiskindofinformationsystemiscalleddistributed
7、informationsystem.ThisgeneralizationisnaturaItodescribethedistributeddatasetsindistributeddataenvironment.Anewconcept.calledmeta—information,isproposedtodescribethedistributedinformationsystemoritssubsystems.Meta-info-rmationeliminatestheredundantdatainthelevelofequivalentclass.anddescri
8、betheintersectionbetweeneverypairofcond}畦onclassanddecisionclassintheformofnumber.弧lus,comparedwithinformationsystem(orsubsystems),meta-informationrequiresfarlessstorage,anditsdatastructureiSmoresimpleandcompact.Duetotheseproperties,recta-informationintegrationgenerallyco
此文档下载收益归作者所有