资源描述：
《Representing nancial time series based on data point importance》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、ARTICLEINPRESSEngineeringApplicationsofArtificialIntelligence21(2008)277–300www.elsevier.com/locate/engappaiRepresentingfinancialtimeseriesbasedondatapointimportancea,b,aabTak-chungFu,Fu-laiChung,RobertLuk,Chak-manNgaDepartmentofComputing,HongKongPolytechnicUnivers
2、ity,Hunghom,Kowloon,HongKongbDepartmentofComputingandInformationManagement,HongKongInstituteofVocationalEducation(ChaiWan),ChaiWan,HongKongReceived12September2006;receivedinrevisedform3March2007;accepted27April2007Availableonline29June2007AbstractRecently,theincre
3、asinguseoftimeseriesdatahasinitiatedvariousresearchanddevelopmentattemptsinthefieldofdataandknowledgemanagement.Timeseriesdataischaracterizedaslargeindatasize,highdimensionalityandupdatecontinuously.Moreover,thetimeseriesdataisalwaysconsideredasawholeinsteadofindiv
4、idualnumericalfields.Indeed,alargesetoftimeseriesdataisfromstockmarket.Stocktimeserieshasitsowncharacteristicsoverothertimeseries.Moreover,dimensionalityreductionisanessentialstepbeforemanytimeseriesanalysisandminingtasks.Forthesereasons,researchispromptedtoaugment
5、existingtechnologiesandbuildnewrepresentationtomanagefinancialtimeseriesdata.Inthispaper,financialtimeseriesisrepresentedaccordingtotheimportanceofthedatapoints.Withtheconceptofdatapointimportance,atreedatastructure,whichsupportsincrementalupdating,isproposedtorepre
6、sentthetimeseriesandanaccessmethodforretrievingthetimeseriesdatapointfromthetree,whichisaccordingtotheirorderofimportance,isintroduced.Thistechniqueiscapabletopresentthetimeseriesindifferentlevelsofdetailandfacilitatemulti-resolutiondimensionalityreductionofthetim
7、eseriesdata.Inthispaper,differentdatapointimportanceevaluationmethods,anewupdatingmethodandtwodimensionalityreductionapproachesareproposedandevaluatedbyaseriesofexperiments.Finally,theapplicationoftheproposedrepresentationonmobileenvironmentisdemonstrated.r2007Els
8、evierLtd.Allrightsreserved.Keywords:Financialtimeseriesrepresentation;Multi-resolutionvisualization;Incrementalupdating;Dimensionalityreduction;Treedata
