當(dāng)代市場調(diào)研 （原第12版）指導(dǎo)手冊 Instructor Manual ch10 市場營銷分析

McDaniel&Gates–MarketingResearch,12thEdition Instructor’sManual

Copyright?2021JohnWiley&Sons,Inc. 10-

CHAPTER10

MarketingAnalytics

LEARNINGOBJECTIVES

1.Understandwhat’sincludedinmarketinganalytics.

2.Reviewtechniquesforanalyzingdata.

3.Gainagreaterunderstandingofbigdata.

4.Exploredatamining.

5.Understanddifferencesinanalyticalforbigandlittledata.

6.Defineartificialintelligence,machinelearninganddeeplearning.

7.Outlinethekeyissuesregardingconsumerprivacy.

KEYTERMS

ArtificialintelligenceorAI

Backpropagation

Behavioraltargeting

Bigdata

CRISP-DMFramework

Datamining

Datavisualization

Deeplearning

Descriptiveanalytics

Machinelearning

Marketinganalytics

Marketingdashboard

Neuralnetworks

Predictiveanalytics

Prescriptiveanalytics

Surgepricing

CHAPTERSUMMARY

Thischapterwilllookatsomeofthetoolsthatenableresearcherstoanalyzeandgaininsightsfromalltypesofdata.Itbeginswithadiscussionofmarketinganalytics,whatitisandwhattheprocessis.Next,itdiscussesbigdata.Thisincludesitsbackground,howitworks,andnowtoanalyzeit.

Next,itdiscussesdescriptive,predictive,andprescriptiveanalytics.Afterthat,itdiscussesdatamining,artificialintelligence,machinelearning,anddeeplearning.Thechapterthistransitionsintobehavioraltargetingandsurgepricing.Next,itdiscussesdatavisualization.Aspartofthatdiscussion,itcoversinfographicsandmarketingdashboards.Itconcludeswithadiscussionofprivacyissues.

QUESTIONSFORREVIEWANDCRITICALTHINKING

Definemarketinganalytics.Whyisitsoimportanttocompanies?

AsdefinedinChapter1,marketinganalyticsisthediscovery,interpretation,andcommunicationofmeaningfulpatternsindata.Thisboilsdowntopredictionorclassificationandtheassociatedinsights.

Marketinganalyticsisimportantbecausecompanieshavetounderstandtheirmarketsinordertoproduceproductsorservicesthataredemandedbytheirmarketsandinordertobeabletorespondtochangesintheirmarket.

Namesometypesofinformationthatmightbefoundinanycompany’sdatabaseandthesourcesofthisinformation.

Anyinformationthefirmcollectsfromitscustomers,suppliers,andothersourcesislikelystoredintheirdatabase.Forexample,Visa,MasterCard,AmericanExpressandothershavemassivedatabaseswhereawiderangeofpurchasesfromretailstores,restaurants,hotels,airlines,onlineretailers,serviceorganizationsandsooncanbeassociatedwithspecificpurchasersaboutwhomthecreditcardcompanieshaveagreatdealofpersonalinformationcoveringage,gender,income,occupation,placeofresidence,andalltheotherinformationyouprovidewhenyoufilloutacreditcardapplication.

Whatismeantbythetermdatamining?Brieflyexplainhowitisdone.

Dataminingisanumbrellatermforanalytictechniquesthatfacilitatefastpatterndiscoveryandmodelbuilding,particularlywithlargedatasets.Thetermislooselyappliedtoanytypeoflarge-scaledataorinformationprocessingaswellasanyapplicationofartificialintelligence,machinelearning,ordeeplearning.Dataminingisperformedusingartificialintelligence,machinelearning,anddeeplearning.

IthasbeensaidthatBigDataanalyticsturnsthescientificmethodonitshead.Whatdoesthismean?

Thescientificmethodisatypeofresearchwhereaproblemisdescribed,relevantdataiscollected,aresearchhypothesis(orhypotheses)isformulated,andthenthehypothesisistestedempirically.Withbigdata,thedataiscollectedfirstandthenanalyzedtofind,notreallyhypotheses,butrathertofindanswerstoquestions.

Whyhasbehavioraltargetingbecomesopopularwithmarketers?Whyisitcontroversial?

Behavioraltargetingistheuseofonlineandofflinedatatounderstandaconsumer’shabits,demographics,andsocialnetworksinordertoincreasetheeffectivenessofonlineadvertising.Thisallowscompaniestoimprovetheirabilitytomarkettotheircustomers.Forexample,Amazonmakingrecommendationstoitscustomers.Behavioraltargetingiscontroversialbecauseoftheprivacyimplicationsandthewayssomeofthedataiscollectedonline.

Whatisdeeplearning?Howisitdifferentfrommachinelearning?HowdotheserelatetoAI?

Machinelearningiswheremachinescanlearnbyexperienceandacquireskillswithouthumaninvolvement.Deeplearningisasubsetofmachinelearningwhereartificialneuralnetworks,algorithmsinspiredbythehumanbrain,learnfromlargeamountsofdataasinmachinelearningbutnowweaddbackpropagationwheremachineslearnfromtheirmistakes.

Bothmachineanddeeplearningareimplementationsofartificialintelligence,wherewecanteachmachinestodothingsthattypicallyrequirehumanintelligence.

Inconnectionwithdeeplearning,whatisbackpropagation?

BackpropagationiswherethedeeplearningAIrealizesithasmadeanerrorandmakesadjustmenttoimprovepredictions.

Whatisdatavisualization?Whyisitimportant?

Datavisualizationconsistsofgraphictoolsthatmakedataunderstandabletoawideraudiencethanjustanalystsanddatascientists.Datavisualizationisimportantbecausehumansunderstanddatamuchquickerandbettervisuallythanbylookingatnumbers.

Whatisamarketingdashboard?Howcanitbeused?

Marketingdashboardsareareportingtoolthatprovidesacomprehensivesnapshotofperformance-basedanalytics,keyperformanceindicators(KPIs),andothermarketingmetrics.Itcanbeusedtovisuallypresentanymarketinginformationcollectedbythefirm.

Dividetheclassintogroupsoffourorfive.EachteamshouldgototheInternetandlookupBigDataanalytics.EachteamshouldthenreporttotheclassonhowaspecificcompanyiseffectivelyusingBigDatatoimprovetheirmarketingefficiency.

Studentresponseswillvary.

REAL-LIFERESEARCH

Case10.1AffiliatedParkingSystemsLookstoNewPricingApproach

KeyPoints

APSownsandoperatesover300parkinglotswithslightlyover33,000parkingplaces.

APShasbeenstruggling,searchingfornewideastoincreaserevenuesfromexistinglots.

APSiswonderingifsurgepricingcouldhelpthem.

APSisinterestedindoingallfeecollectionfromtotallyelectronicallytofurtherreducevariablecosts.

APSwantstovarypricingbasedonthelevelofdemandforparkinginrealtime.

Questions

WouldyousaythatBillisontherighttrackregardingtheneedforartificialintelligencetoimplementdynamicpricing?Whydoyousaythat?

Studentopinionswillvary.However,withafixedinventoryofparkingspaces,dynamicpricingoffersabouttheonlyoptiontheyhaveforincreasingrevenue.

Ifheweretopursuetheideafurther(heobviouslywouldneedhelpfromaconsultingfirm),whatdatawouldbeneededtoimplementsurgeordynamicpricing?

Sincetheyown300parkinglots,thisistheperfectopportunitytopilottest(e.g.testmarket)theconceptifthatisdesired.

Inordertoimplementsurgepricing,theywouldneedtoknowhowdemandvarieswithtime-of-day,day-of-week,andspecialevents.Theycouldbeginbycollectingdemandfromtheelectronicsystemsandattendants.Somelotshavemanualsystemsandthesewouldbeincompatiblewithbothdatacollectionandsurgepricingsotheywouldneedtobeupgradedforthesystemtowork.

Wouldmodelsbeneeded?Whatwouldthemodelsdo?Howmighttheybedeveloped?

Machinelearningwouldberequiredtomodeldemandandadjustpricingonanongoingbases,raisingpriceswithspacesareinhighdemandandloweringpriceswhenspacesareinlowdemand.

Describetheultimatesystemthatwouldbeneededintermsofinputsneeded,howthoseinputswouldbecaptured,modelsneeded(justageneralsenseofwhatthemodelswouldneedtodo),howpricingwouldbecommunicatedtoperspectiveusersandhowfeeswouldbecollected.Mappingitalloutinadiagramwithafewcommentsonwhatisoccurringateachstepisprobablyagoodapproachtoansweringthisquestion.

WhilethisinitiallysoundssimilartosurgepricingwithUber,itisactuallyverydifferent.WithUber,youagreeonthepriceaheado