交通数据挖掘技术（Data Mining for Transportation）_东南大学中国大学mooc章节课后测试答案期末考试题库2024年.docxVIP

交通数据挖掘技术（DataMiningforTransportation）_东南大学中国大学mooc章节课后测试答案期末考试题库2024年

Neutralnetworkclassifierslikebackpropagationcanbeusedasclassifiersinthreshold-movingapproach.

答案:正确

Theprinciplethreshold-movingistomovethedecisionthreshold,sothattherareclasstuplesareeasiertoclassify.

答案:正确

ForestRIcreatesnewattributesthatarealinearcombinationoftheexistingattributes.

答案:错误

Inrandomforest,usingarandomselectionofattributesateachnodetodeterminethesplit.

答案:正确

Togeneratethesingleclassifier,differentmodelmaybeusedtodealwithdifferentdatasubset.

答案:正确

EnsemblemethodcannotparallelizablebecausenoteverybaseclassifiercanbeallocatedtoadifferentCPU.

答案:错误

Ensemblestendtoyieldbetterresultswhenthereisasignificantdiversityamongthebasemodels.

答案:正确

Whichstepisnecessarywhenconstructinganensemblemodel?

答案:Creatingmultipledataset_Constructingasetofclassifiersfromthetrainingdata_Combiningpredictionsmadebymultipleclassifierstoobtainfinalclasslabel

Whichoneisrightwhendealingwiththeclass-imbalanceproblem？

答案:Smotealgorithmaddssynthetictuplesthatareclosetotheminoritytuplesintuplespace._Threshold-movingandensemblemethodswereempiricallyobservedtooutperformoversamplingandundersampling.

Howtodealwithimbalanceddatain2-classclassification?

答案:Oversampling_Undersampling_Threshold-moving_Ensembletechniques

Anoutlierisadataobjectthatdeviatessignificantlyfromtherestoftheobjects,asifitweregeneratedbyadifferentmechanism.

答案:正确

MahalanobisDistanceaccountsfortherelativedispersionsandinherentcorrelationsamongvectorelements,whichisdifferentfromEuclideanDistance.

答案:正确

Whenidentifyingoutlierswithadiscordancytest,thedatapointisconsideredasanoutlierifitfallswithintheconfidenceinterval.

答案:错误

Statistic-basedmethodneedstorequireknowingthedistributionofthedataandthedistribution