Part 9 Review Questions and Exercises.docx

Part 9 Review Questions and Exercises.docx

《Part 9 Review Questions and Exercises.docx》由会员分享，可在线阅读，更多相关《Part 9 Review Questions and Exercises.docx（13页珍藏版）》请在冰豆网上搜索。

Part9ReviewQuestionsandExercises

SOLUTIONSTOREVIEWQUESTIONS

ANDEXERCISES

FORPART9–BUSINESSINTELLIGENCE（CHAPTERS32–35）

SolutionstoReviewQuestionsandExercises

Chapter32DataWarehousingConcepts3

Chapter33DataWarehousing-Design6

Chapter34OLAP8

Chapter35DataMining13

Chapter32DataWarehousingConcepts

ReviewQuestions

32.1Describewhatismeantbythefollowingterms,whendescribingthecharacteristicsofthedatainadatawarehouse:

（a）subject-oriented;

（b）integrated;

（c）time-variant;

（d）non-volatile.

SeeSection32.1.2.

32.2DiscusshowOnlineTransactionProcessing（OLTP）systemsdifferfromdatawarehousingsystems.

SeeSection32.1.4

32.3Discussthemainbenefitsandproblemsassociatedwithdatawarehousing.

ForthemainbenefitsofdatawarehousingseeSection32.1.3andforthemainproblemsassociatedwithdatawarehousingseeSection32.1.5.

32.4Presentadiagrammaticrepresentationofthetypicalarchitectureandmaincomponentsofadatawarehouse.

ForadiagramofthetypicalarchitectureofadatawarehouseseeFigure32.1.

32.5Describethecharacteristicsandmainfunctionsofthefollowingcomponentsofadatawarehouse.

（a）loadmanagerSeeSection32.2.3

（b）warehousemanagerSeeSection32.2.4

（c）querymanagerSeeSection32.2.5

（d）metadataSeeSection32.2.9

（e）end-useraccesstools.SeeSection32.2.10

32.6Describetheprocessesassociatedwithdataextraction,cleansing,andtransformationtools.

TheextractionsteptargetsoneormoredatasourcesfortheEDW;thesesourcestypicallyincludeOLTPdatabasesbutcanalsoincludesourcessuchaspersonaldatabasesandspreadsheets,enterpriseresourceplanning（ERP）files,andwebusagelogfiles.Thedatasourcesarenormallyinternalbutcanalsoincludeexternalsources,suchasthesystemsusedbysuppliersand/orcustomers.

Thetransformationstepappliesaseriesofrulesorfunctionstotheextracteddata,whichdetermineshowthedatawillbeusedforanalysisandcaninvolvetransformationssuchasdatasummations,dataencoding,datamerging,datasplitting,datacalculations,andcreationofsurrogatekeys.Theoutputfromthetransformationsisdatathatiscleanandconsistentwiththedataalreadyheldinthewarehouse,andfurthermore,isinaformthatisreadyforanalysisbyusersofthewarehouse.

Theloadingofthedataintothewarehousecanoccurafteralltransformationshavetakenplaceoraspartofthetransformationprocessing.Asthedataloadsintothewarehouse,additionalconstraintsdefinedinthedatabaseschemaaswellasintriggersactivatedupondataloadingwillbeapplied（suchasuniqueness,referentialintegrity,andmandatoryfields）,whichalsocontributetotheoveralldataqualityperformanceoftheETLprocess.

32.7Describethespecializedrequirementsofarelationaldatabasemanagementsystem（RDBMS）suitableforuseinadatawarehouseenvironment.

SeeSection32.4.2

32.8Discusshowparalleltechnologiescansupporttherequirementsofthedatawarehouse.

SeelasttopicdiscussedinSection32.4.2undertheheadingParalleldatabasetechnologies.

32.9Discusstheimportanceofmanagingmeta-dataandhowthisrelatestotheintegrationofthedatawarehouse.

SeeSection32.4.3.

32.10Discussthemaintasksassociatedwiththeadministrationandmanagementofadatawarehouse.

Thedatawarehouseadministrationandmanagementtoolsmustbecapableofsupportingthefollowingtasks:

∙monitoringdataloadingfrommultiplesources;

∙dataqualityandintegritychecks;

∙managingandupdatingmeta-data;

∙monitoringdatabaseperformancetoensureefficientqueryresponsetimesandresourceutilization;

∙auditingdatawarehouseusagetoprovideuserchargebackinformation;

∙replicating,subsetting,anddistributingdata;

∙maintainingefficientdatastoragemanagement;

∙purgingdata;

∙archivingandbacking-updata;

∙implementingrecoveryfollowingfailure;

∙securitymanagement.

SeeSection32.4.4.

32.11Discusshowdatamartsdifferfromdatawarehousesanddiscussthemainreasonsforimplementingadatamart.

ForadiscussiononhowdatamartsdifferfromdatawarehousesseeintroductoryparagraphsofSection321.5andforreasonsforimplementingadatamartseeSection32.5.1.

32.12DescribethefeaturesofOraclethatsupportthecorerequirementsofdatawarehousing.

SeeSection32.6.

Exercises

32.13YouareaskedbytheManagingDirectorofDreamHometoinvestigateandreportontheapplicabilityofdatawarehousingfortheorganization.ThereportshouldcomparedatawarehousetechnologywithOLTPsystemsandshouldidentifytheadvantagesanddisadvantages,andanyproblemareasassociatedwithimplementingadatawarehouse.ThereportshouldreachafullyjustifiedsetofconclusionsontheapplicabilityofadatawarehouseforDreamHome.

Theformatandtheappropriatecontenttobecoveredinansweringthisquestionisdescribedinthequestionset.

Chapter33DataWarehousing-Design

ReviewQuestions

33.1Discusstheactivitiesassociatedwithinitiatinganenterprisedatawarehouse（EDW）project.

Tobeginadatawarehouseproject,weneedanswersforquestionssuchas:

whichuserrequirementsaremostimportantandwhichdatashouldbeconsideredfirst?

Also,shouldtheprojectbescaleddownintosomethingmoremanageable,yetatthesametimeprovideaninfrastructurecapableofultimatelydeliveringafull-scaleenterprise-widedatawarehouse?

TherequirementscollectionandanalysisstageofanEDWprojectinvolvesinterviewingappropriatemembersofstaffsuchasmarketingusers,financeusers,salesusers,operationalusers,andmanagementtoenabletheidentificationofaprioritizedsetofrequirementsfortheenterprisethatthedatawarehousemustmeet.Atthesametime,interviewsareconductedwithmembersofstaffresponsibleforOLTPsystemstoidentify,whichdatasourcescanprovideclean,valid,andconsistentdatathatwillremainsupportedoverthenextfewyears.

Theinterviewsprovidethenecessaryinformationforthetop-downview（userrequirements）andthebottom-upview（whichdatasourcesareavailable）oftheEDW.

33.2CompareandcontrasttheapproachestakeninthedevelopmentofanEDWbyInmon’sCorporateInformationFactory（CIF）andKimball’sBusinessDimensionalLifecycle.

Inmon’sapproachistostartbycreatingadatamodelofalltheenterprise’sdata;oncecomplete,itisusedtoimplementanEDW.TheEDWisthenusedtofeeddepartmentaldatabases（datamarts）,whichexisttomeettheparticularinformationrequirementsofeachdepartment.TheEDWcanalsoprovidedatatootherspecializeddecisionsupportapplicationssuchasCustomerRelationshipManagement（CRM）.Inmon’smethodologyusestraditionaldatabasemethodsandtechniquestodeveloptheEDW.Forexample,entity–relationship（ER）modeling（Chapter12）isusedtodescribetheEDWdatabase,whichholdstablesthatareinthirdnormalform（Chapter14）.InmonbelievesthatafullynormalizedEDWisrequiredtoprovidethenecessaryflexibilitytosupportthevariousoverlappinganddistinctinformationrequirementsofallpartsoftheenterprise.

Kimball’sapproachusesnewmethodsandtechniquesinthedevelopmentofanEDW.Kimballstartsbyidentifyingtheinformationrequirements（referredtoasanalyticalthemes）andassociatedbusinessprocessesoftheenterprise.ThisactivityresultsinthecreationofacriticaldocumentcalledaDataWarehouseBusMatrix.Thematrixlistsallofthekeybusinessprocessesofanenterprisetogetherwithanindicationofhowtheseprocessesaretobeanalyzed.Thematrixisusedtofacilitatetheselectionanddevelopmentofthefirstdatabase（datamart）tomeettheinformationrequirementsofaparticulargroupofusersoftheenterprise.Thisfirstdatamartiscriticalinsettingthesceneforthelaterintegrationofotherdatamartsastheycomeonline.TheintegrationofdatamartsultimatelyleadstothedevelopmentofanEDW.Kimballusesanewtechniquecalleddimensionalitymodelingtoestablishthedatamodel（referredtoasadimensionalmodel（DM）foreachdatamart.Dimensionalitymodelingresultsinthecreationofadimensionalmodel（commonlycalledastarschema）foreachdatamartthatishighlydenormalized.Kimballbelievesthattheuseofstarschemasisamoreintuitivewaytomodeldecisionsupportdataandfurthermorecanenhanceperformanceforcomplexanalyticalqueries.

33.3DiscussthemainprinciplesandstagesassociatedwithKimball’sBusinessDimensionalLifecycle.

ThemainstagesaresummarizedinFigure33.1.

33.4Discusstheconceptsassociatedwithdimensionalitymodeling.

Everydimensionalmodel（DM）iscomposedofonetablewithacompositeprimarykey,calledthefacttable,andasetofsmallertables,calleddimensiontables.Eachdimensiontablehasasimple（noncomposite）primarykeythatcorrespondsexactlytooneofthecomponentsofthecompositekeyinthefacttable.Inotherwords,theprimarykeyofthefacttableismadeupoftwoormoreforeignkeys.Thischaracteristic“star-like”structureiscalledastarschemaorstarjoin.AnotherimportantfeatureofaDMisthatallnaturalkeysarereplacedwithsurrogatekeys.Thismeansthateveryjoinbetweenfactanddimensiontablesisbasedonsurrogatekeys,notnaturalkeys.Eachsurrogatekeyshouldhaveageneralizedstructurebasedonsimpleintegers.TheuseofsurrogatekeysallowsthedatainthewarehousetohavesomeindependencefromthedatausedandproducedbytheOLTPsystems.

33.5Describehowstar,snowflake,andstarflakeschemasdiffer.

Starschemaisalogicalstructurethathasafacttablecontainingfactualdatainthecenter,surroundedbydim