当代市场调研10E全套配套课件中英文PPTch18.ppt 24页

ChapterEighteen18-1 MultivariateAnalysisMultivariateRegressionAnalysisMultipleDiscriminantAnalysisClusterAnalysisFactorAnalysisConjointAnalysisAgeneraltermforstatisticalproceduresthatsimultaneouslyanalyzemultiplemeasurementsoneachindividualorobjectunderstudy.18-2KeyTerms&Definitions MultivariateAnalysis18-3KeyTerms&Definitions MultipleRegressionKeyConceptsCoefficientofDetermination:Measuredchangesinthedependentandindependentvariables.RegressionCoefficients:Effectoftheindependentvariableonthedependentvariable.DummyVariables:They"renominallyscaledvariablesincludedinregressionanalysis.Aprocedureforpredictingthelevelormagnitudeofa(metric)dependentvariablebasedonthelevelsofmultipleindependentvariables.18-4KeyTerms&Definitions MultipleRegressionAnalysis18-5KeyTerms&Definitions MultipleRegressionAnalysis18-6KeyTerms&Definitions UsesofMultipleRegressionAnalysis18-7KeyTerms&Definitions CoefficientofDetermination:Measureofthepercentageofthevariationinthedependentvariableexplainedbyvariationsintheindependentvariables.RegressionCoefficients:Estimatesoftheeffectofindividualindependentvariablesonthedependentvariable.DummyVariables:Inregressionanalysis,awayofrepresentingtwo-groupordichotomous,nominallyscaledindependentvariablesbycodingonegroupas0andtheotheras1.18-8KeyTerms&DefinitionsMultipleRegressionMeasurementApplications PotentialUseandInterpretationProblemsCollinearity:Correlationofindependentvariableswitheachother,whichcanbiasestimatesofregressioncoefficients.Causation:Inferencethatachangeinonevariableisresponsiblefor(caused)anobservedchangeinanothervariable.ScalingofCoefficients:Amethodofdirectlycomparingthemagnitudesoftheregressioncoefficientsofindependentvariablesbyscalingtheminthesameunitsorbystandardizingthedata.18-9KeyTerms&Definitions StandardizationProcess18-10KeyTerms&Definitions MultipleDiscriminantAnalysisMultipleDiscriminantAnalysis:Procedureforpredictinggroupmembershipfora(nominalorcategorical)dependentvariableonthebasisoftwoormoreindependentvariables.MetricScale:Atypeofquantitativethatprovidesthemostprecisemeasurement.NominalorCategorical:Atypeofnon-metricqualitativedatascalethatonlyusesnumberstoindicatemembershipinagroup(e.g.,1=male,2=female).Mostmathematicalandstatisticalprocedurescannotbeappliedtonominaldata.18-11KeyTerms&Definitions MultipleDiscriminantAnalysisDiscriminantScore:Scorethatisthebasisforpredictingtowhichgroupaparticularobjectorindividualbelongs;alsocalledZscore.DiscriminantCoefficient:Estimateofthediscriminatorypowerofaparticularindependentvariable;alsocalleddiscriminantweight.ClassificationMatrix:Amatrixortablethatshowsthepercentagesofpeopleorthingscorrectlyandincorrectlyclassifiedbythediscriminantmodel.18-12KeyTerms&Definitions ClusterAnalysisThegeneraltermforstatisticalproceduresthatclassifyobjects,orpeople,intosomenumberofmutuallyexclusiveandexhaustivegroupsonthebasisoftwoormoreclassificationvariables.18-13KeyTerms&DefinitionsConsumerswhofrequentlyeatoutbutseldomeatatfast-foodrestaurants.Peoplewhofrequentlyeatoutandalsofrequentlyeatatfast-foodrestaurants.Peoplewhodonotfrequentlyeatoutorfrequentlyeatatfast-foodrestaurants. ExampleofClusterAnalysisK-meansClusterAnalysis:Multivariateprocedurethatgroupspeopleorobjectsintoclustersbasedontheirclosenesstoeachotheronanumberofvariables.18-14KeyTerms&Definitions FactorAnalysisAprocedureforsimplifyingdatabyreducingalargesetofvariablestoasmallersetoffactorsofcompositevariablesbyidentifyingdimensionsofthedata.Factor:Alinearcombinationofvariablesthatarecorrelatedwitheachother.18-15KeyTerms&Definitions FactorScoresInfactoranalysis,afactorscoreiscalculatedoneachfactorforeachsubjectinthedataset.Forexample,inafactoranalysiswithtwofactors,thefollowingequationsmightbeusedtodeterminefactorscores:18-16KeyTerms&Definitions FactorLoadingCorrelationbetweenfactorscoresandtheoriginalvariables.OtherKeyIssues:NamingFactorsNumberoffactorstoRetain18-17KeyTerms&Definitions ConjointAnalysisMultivariateprocedureusedtoquantifythevalueconsumersassociatewithdifferentlevelsofproduct/serviceattributesorfeatures.18-18KeyTerms&Definitions ConjointAnalysis–EstimatingUtilitiesUtilities:Therelativevalueofattributelevelsdeterminedthroughconjointanalysis.18-19KeyTerms&Definitions ConjointAnalysis–EstimatingUtilitiesUtilities:Therelativevalueofattributelevelsdeterminedthroughconjointanalysis.18-20KeyTerms&Definitions LimitationsofConjointAnalysis18-21KeyTerms&DefinitionsThereisacertaindegreeofartificiality.Respondentsmaybemoredeliberateintheirchoiceprocesses.Thesurveymayprovidemoreproductinformationthanrespondentswouldgetinarealmarketsituation.Theadvertisingandpromotionofanynewproductcanleadtoconsumerperceptionsthatareverydifferentfromthosecreatedviadescriptionsusedinasurvey. AftersomePre-processing,theactualdataminingprocessismadeofthreesteps:1.Modeling—byclustering,classification,orestimation2.Validation3.ApplicationoftheResultsPredictiveAnalyticsProcessKeyTerms&Definitions18-22 Examplesofeachoneintoday’sworld:PrivacyLegalEthicsDataMiningConcerns18-23KeyTerms&Definitions 18-24KeyTerms&DefinitionsMultivariateAnalysisMultipleRegressionAnalysisCoefficientofDeterminationRegressionCoefficientsDummyVariablesCollinearityCausationScalingofCoefficientsMultipleDiscriminantAnalysisMetricScaleNominalorCategoricalDiscriminantScoreLinksandbuttonareactivewhenin“SlideShowMode”KeyTerms&DefinitionsDiscriminantCoefficientClassificationMatrixClusterAnalysisK-meansClusterAnalysisFactorAnalysisFactorScoresFactorLoadingConjointAnalysisLimitationsofConjointAnalysisEstimatingUtilities