时间序列时序关联规则挖掘研究(6)
发布时间:2021-06-06
发布时间:2021-06-06
一一
ABSTRACT些兰里坠璺
Temporalassociationrulesoftimeseriesalethetemporalconstrainingassociationamongpartlychangesoftimeseries.Partlychangesoftimeseriesthemselveshavetimesequence,SOtimeorderiSacharacteristicoftheassociation.Timeserieshavethecharacteristicsofdatadensenessandstochasticfluctuation,andtemporalassociationrulesofpartly
SOchangesareimpliedinthelargedataset,therulescallbeobtainedonlythroughdatamining.
isaTheminingoftemporalassociationrulesoftimeseries
engineering,which
datasystematicCanbedividedintotimeseriesdatapre-processing,timeseriesseriesdatasimilaritymeasure,therequirementoftemporalcompression,time
associationrulesandtheinterpretationandevaluationoftemporalassociation
arules.Theresearch
lot,butisfarfromonminingmethodsoftemporalassociationruleshasgainedmainpointsareaspeffection.Thefollows.
statisticsishardtOgainthe
on(1)Inrecognizingoutlier,themethodbasedonsample’Sdistributionparameter,themethodbased
changetheauthenticityoforiginaltime
ratiohasawavelettransformwillonseries,andthemethodbasedlikelihoodlargeamountofcalculation.
given
enduplengthwith(2)Inandminingtheclassicaltemporalassociationrules,thetimeseriesarediscreditedintosequentialpatternsbytheslidingwindowwiththesteps.Thefrequentpatternwillbeacquiredanditwill
strengthenedtemporalassociationrules.Becausethelengthandstepfortheslidingwindowarearbitrary,thereiSalotofuncertaintyintheresultfromthetimeseriescompression.
(3)Similarity
patterninmeasureoftimeseriesisthebaseforacquiringthefrequentsequentialpatterns,andalsodecidestheobtainmentoftemporalassociationrules.Themeta-patternmonotony
distancebothhavesomeflawsin
measureofmeta—patternhasdistanceandthemeta.patternvectordefiningthemeta-pattern,SOproblems.Andthethesimilaritymethodsofsomeexisting
下一篇:哲学选择题训练