LEARNING MULTIVARIATE TIME SERIES CAUSAL GRAPHS BASED ON CONDITIONAL MUTUAL INFORMATION

(整期优先)网络出版时间:2013-01-11
/ 1
Detectionandclarificationofcause-effectrelationshipsamongvariablesisanimportantproblemintimeseriesanalysis.Thispaperprovidesamethodthatemploysbothmutualinformationandconditionalmutualinformationtoidentifythecausalstructureofmultivariatetimeseriescausalgraphicalmodels.Athree-stepprocedureisdevelopedtolearnthecontemporaneousandthelaggedcausalrelationshipsoftimeseriescausalgraphs.Contrarytoconventionalconstraint-basedalgorithm,theproposedalgorithmdoesnotinvolveanyspecialkindsofdistributionandisnonparametric.Thesepropertiesareespeciallyappealingforinferenceoftimeseriescausalgraphswhenthepriorknowledgeaboutthedatamodelisnotavailable.Simulationsandcaseanalysisdemonstratetheeffectivenessofthemethod.