简介：Evaluationofthecyclicshearmodulusofsoilsisacrucialbutchallengingtaskformanygeotechnicalearthquakeengineeringandsoildynamicissues.Improperdeterminationofthispropertyunnecessarilydrivesupdesignandmaintenancecostsorevenleadstotheconstructionofunsafestructures.Duetothecomplexitiesinvolvedinthedirectmeasurement,empiricalcurvesforestimatingthecyclicshearmodulushavebeencommonlyadoptedinpracticeforsimplicityandeconomicalconsiderations.However,asystematicandrobustapproachforformulatingareliablemodelandempiricalcurveforcyclicshearmoduluspredictionforclayeysoilsisstilllacking.Inthisstudy,theBayesianmodelclassselectionapproachisutilizedtoidentifythemostsignificantsoilparametersaffectingthenormalizedcyclicshearmodulusandareliablepredictivemodelfornormallytomoderatelyover-consolidatedclaysisproposed.Resultsshowthatthepredictabilityandreliabilityoftheproposedmodeloutperformsthewell-knownempiricalmodels.Finally,anewdesignchartisestablishedforpracticalusage.