不完备信息系统中优势关系粗糙集决策方法-研究

《不完备信息系统中优势关系粗糙集决策方法-研究》由会员上传分享，免费在线阅读，更多相关内容在应用文档-天天文库。

1、哈尔滨理工大学管理学硕士学位论文StudyDecision—makingapproachesBasedonDominanceRelationRoughSetsinIncompleteInformationSystemAbstractClassicalroughsettheoryisbasedonequivalencerelation，assumingthateachsampleobjectshavedefiniteattributevalues．Butintheactualdataminingproblems，thedatatobeprocessedisalways

2、incompletebecauseoftheinfluenceofvariousfactors．whichwillaffectthedataanalysis．Therefore，ithasbecomeoneoftheimportantresearchsubjectsthatroughsettheoryisappliedtoacquireknowledgefromincompleteinformationsystematpresent．Dominancerelationroughsettheoryhasbeenextendedbymanyrelatedscholars

3、，whichisappliedintheincompleteinformationsystem．Thispaperisaimedatfurtherstudyingallkindsofextendeddominancerelationroughsetinincompleteinformationsystem．Comparativeanalysisoftheexistingdominancerelationroughsetmodel．Foundthatthemodelstillexistsomedisadvantages：theexistingdominancerela

4、tionroughsetdirectruledoutthemwithoutconsiderationthatattributevaluesoftwoobjectswithsameattributeareallnulloroneofthetwoobject’SattributevalueiSnulIandtheotheroneiSthemiddlevalue(TheseattributesiScalledunablecomparedattribute)．Thedecisionresultswillhaveagreatdeviationwiththeactualresu

5、ltthattheexistingdominancerelationroughsetiSappliedtomakedecisionswhentheabovesituationOccursfrequently．Accordingtotheaboveproblem，thispaperintroducedtheconceptofthecriticalvalueinorderthatincompleteinformationsystemisdividedintotwocasestodiscussaccordingtothenumberofunablecomparedattr

6、ibutes．Thefirstcase，buildingtheimp·rovedandlimiteddominancerelationroughsetmodelwhichretainstheadvantagesoflimiteddominancerelationundertheconditionthatthecriticalvalueisgreaterthanorequaltothecriticalvalue．Thesecondcase，makingapairedcomparisontablewhoseattributevalueisintervalnumber，．

7、II．Buildingavariableprecisionanddominancerelationroughsetmodelcomblnetheexpansiontheoriesofdominancerelationroughsetandvariablepreclslonroughset，givingmethodsofattributereductionandextractionofdeclslonrules,settingtheruleconfidencethresholdvalueofthevariablepreclstonroughset(VPRS)isv