粗糙集理论在挖掘周期性频繁模式中的应用(IJITCS-V8-N7-8)

I.J.InformationTechnologyandComputerScience,2016,7,53-60PublishedOnlineJuly2016inMECS()DOI:10.5815/ijitcs.2016.07.08Copyright©2016MECSI.J.InformationTechnologyandComputerScience,2016,7,53-60OntheUseofRoughSetTheoryforMiningPeriodicFrequentPatternsManjeetSamoliyaDepartmentofCSE&IT,MadhavInstituteofTechnologyandScience,Gwalior(M.P),474005,IndiaE-mail:[e-mail]entofCSE&IT,MadhavInstituteofTechnologyandScience,Gwalior(M.P),474005,IndiaE-mail:[e-mail]—ThispaperpresentsanewAprioribasedapproachforminingperiodicfrequentpatternsfromthetemporaldatabase.Theproposedapproachutilizestheconceptofroughsettheoryforobtainingreducedrepresentationoftheinitiallyconsideredtemporaldatabase.Inordertoconsideronlytherelevantitemsforanalyzingseasonaleffects,adecisionattributefestivalhasbeenconsidered.Ithasbeenobservedthattheproposedapproachworksfinefortheanalysisoftheseasonalimpactonbuyingbehaviorofcustomers.Consideringthecapabilityofapproachfortheanalysisofseasonalprofitabilityconcern,decisionmaking,andfuturemarketingmayuseitfortheimportantdecision-makingprocessfortheupliftingofsell.IndexTerms—Associationrulemining,frequentpattern,periodicpatternmining,roughsets,temporaldatabase.I.INTRODUCTIONDataminingreferstoextractingusefulandvaluableinformationfromthehugeamountofdata[1].Byperforminginformationmining,interestinglearning,re-liabilities,orhigh-leveldatacanberemovedfromthedatabaseandviewedorbrowsedfromdiversemethodologies.Thediscoveredlearningcanbeconnectedtochoicemaking,procedurecontrol,dataadministration,andqueryhandling.Informationminingisviewedasastandoutamongstthemostimperativewildernessesindatabaseframeworksandoneofthepromisinginterdisciplinaryadvancementsinthedatabusiness.Informationmining,withitsguaranteetoproductivelyfindprofitable,non-cleardatafromexpansivedatabases[2],isespeciallypowerlesstoabuse.So,theremaybeavariancebetweeninformationminingandprotection.Quickadvancesininformationgatheringandstorageinnovationhaveenabledtheassociationtocollectinconceivablemeasuresofinformation.Ithasbeenassessedthattheamountofinformationontheplanetcopiesevery20monthsandthesizeandnumberofthedatabasesareexpandingconsiderablyquicker.However,extractinghelpfuldatahasdemonstratedgreatlydifficult.Conventionalinvestigationdevicesandstrategiescan'tbeutilizedduetothemassivesizeofinformation[3].Thefieldofdataminingisvastandconsistsofvarioussub-fieldswhichareavailableintheliterature.Byusingdatamining,wecandiscoverthepatternsofdata.Therearedataminingteamsworkinginbusiness,government,financialservices,biology,medicine,riskandintelligence,science,andengineering.Wheneverwecollectdata,dataminingisappliedandprovidingnewknowledgeintohumanendeavor.Oneofthemajorgrowingfieldsisthefrequentpatternmininginassociationruleminingwhichhasbeenexplainedindetailfurther.A.AssociationRuleMiningAssociationruleminingisawell-knownmethodtodiscoverinterestingrulesandrelationsbetweenvariousitemsinlargedatabases.Basedontheconceptofstrongrules,RakeshAgrawaletal.[4]introducedassociationrulesforfindingregularitiesbetweenitemsinlarge-scaleexchangeinformationrecorded.Anassociationrulehastwodifferentsections,anantecedent(if)andalsoaconsequent(then).Anantecedentisacreateditemintheknowledge.Aconsequenceisafounditemcombinationwiththeantecedent.Associationrulesarecompletethroughexaminingdataforfrequentpatternsif/thenandusingthesupportandconfidencefunctiontoidentifythemostimportantrelationships.Supportshowsthatfrequentitemsshowuphowmanytimesinadatabase.Support(s)ofanassociationruleisdefinedasthepercentage/fractionofrecordsthatcontainX∪Ytothetotalnumberofrecordsinthedatabase.ThecountisincreasedwhenitemisdifferenttransactionTindatabaseDduringthescanningprocess.Itmeanssupportcountdoesn’ttakeintoconsiderationquantityoftheitemintoaccount.Confidencedemonstratesthetimesquantitytheif/thenstatementsdiscoveredtobevalid[14].Confidenceofanassociationruleisdefinedasthepercentage/fractionofthenumberoftransactionsthatcontainX∪YtothetotalnumberofrecordsthatcontainX,whereifthepercentageexceedsthethresholdofconfidenceaninterestingassociationruleX→Ycanbegenerated.Associationrulegenerationisgenerallydividedintotwostepprocess.54OntheUseofRoughSetTheoryforMiningPeriodicFrequentPatternsCopyright©2016MECSI.J.InformationTechnologyandComputerScience,2016,7,53-60Aminimumsupportthresholdisappliedtofindallfrequentitem-setsinadatabase.Aminimumconfidenceconstraintisappliedtothesefrequentitem-setsinordertoformrules.NYXYXSupport.(1)XYXYXConfidence.(2)Numerousbusinessenterprisescollectvastamountsofinformationfromtheirregularoperations.Forexample,largequantitiesofclientpurchaseinformationaregathereddaybydayatthegrocerystorecheckoutcounters.Associationruleminingcanbeofmanyformsandvariousotherfieldscanbeeasilymergedinthis.Associationruleminingdependsmainlyuponfindingthefrequentitemsets.ForfindingthefrequentitemsetweuseApriorialgorithmasoneofthemostimportantalgorithmtofindthefrequentitemset.B.AprioriAlgorithmApriorialgorithmwa