Continuous Data Quality ManagementWord下载.docx
- 文档编号:15813137
- 上传时间:2022-11-16
- 格式:DOCX
- 页数:9
- 大小:23.90KB
Continuous Data Quality ManagementWord下载.docx
《Continuous Data Quality ManagementWord下载.docx》由会员分享,可在线阅读,更多相关《Continuous Data Quality ManagementWord下载.docx(9页珍藏版)》请在冰豆网上搜索。
NomatterhowwellanenterpriseimplementsaCRM,ERP,SCM,BusinessIntelligence,orDataWarehouseproject,poordataqualitycandestroyitsutilityandcostrealdollars.Accordingtorecentindustrystudies:
Poordataqualitycostsbusinesses$611billionperyearintheUnitedStatesalone(TDWI).
75%ofbusinesseshaveexperiencedsignificantproblemsduetofaultydata(PWC).
Only33%ofbusinessesfeltconfidentinthequalityoftheircompany'
sdata.
Nowimaginethedownstreamimpactofthissamepoorqualitydatafuelingbusinessdecisions.Notonlywillbaddecisionbemade,butwhenemployeesandmanagersstoptrustingtheirbusinessintelligenceapplications,theentireinvestmentinthesecostlysystemscanbejeopardized.It’squitesimple:
businessesthatignoredataqualitydosoattheirownperilandexpense.
NoTimeToThink
Zero-latencydecision-making,basedonreal-time/near-timemonitoringofbusinessactivity,isacleartrendinthebusinesscommunity.Toenablethisinstantaccess/instantresponsetotakeplace,ablendingofreal-timebusinessintelligencewithhistoricaldataisrequired,creatingatremendousdependencyoncontinuousdataquality.
Traditionally,large-scaleapplicationssuchasERP,CRM,BusinessIntelligence,ProcessManagement,Middleware,andDataIntegrationhaveexperienceddifficultieswithdataqualityproblems.ERPusuallyservesasadatacollectionpointwheredataerrorscanbeintroducedintoenterprisedataassets.CRMtracksanever-changing,ever-movingtargetofsalesandcustomers,wherepoordataqualitycanresultinmissedsalesopportunitiesaswellasupsetcustomers.BusinessIntelligencecanonlyprovideinformationasaccurateasthesourcedata.BusinessProcessManagementisonlyeffectivewhenthedataexchangedisaccurate.MiddlewareandDataIntegrationmerelymovefaultydatafrompointAtopointB.Whileanimplementationmayincorporatedatacleansingaspartofaproject,ongoingmonitoringofdataqualityafterhandoveriscurrentlynotatypicalITassignment.
Zero-latencydecision-makingthatdoesnotpaycloseattentiontodataqualityislikelytofail.Unfortunately,thetraditionalapproachtodataqualityissimplynotrobustenoughtomeetthesenewdemands.Dataqualitysolutionshavetraditionallyclusteredaroundnamestandardization,addresshygiene,anddemographicaccuracy.Whilethesepropertiesofdataqualityareimportant,theyarepracticallyuselessinanearzero-latencyimplementation.Traditionaldataqualitytoolsareinherentlybatch-orientedandsingle-shotendeavors,whereasthecurrentneedisforatransactional,real-timesolution.Evenifthetechnologicalhurdlesforreal-timenamestandardization,addresscleansing,anddemographicassignmentcouldbeovercome,aframeworkhasnotexistedformeasuringtheimpactoftheseactivities.Eveniftheycouldbecleanedup,thesamemistakeswouldcontinuetooccurwhendataisaddedormanipulated,requiringevenmorecleanup.
Infact,dataqualitygoesfarbeyonddatacleansing.Dataqualityindicateshowwellenterprisedatamatchesupwiththerealworldatanygiventime.Ifdecision-makingisbaseduponpoorqualitydata,thenbydefinitionthereisnoabilitytomakeaccuratedecisions.Luckily,thereareseveralindicatorsfordataqualitythatcanbeusedtodefineandmeasurethestateofdataqualityintheenterprise.
DefiningDataQuality
DataQualitycanbebrokenintothefollowingeightcategories:
●Domains
●Definitions
●Completeness
●Validity
●Businessrules
●Structuralintegrity
●Transformations
●Dataflows
●Domains
Domainsdescribetherangesandtypesofvaluespresentinadataset.Thetypicalerrorsthatcanoccurrelatingtodomainsare:
Unexpecteddomainvalues.Thedocumentationforasystemindicatesthatthevaluesforacolumnwillbe(A,B,C),butthedataactuallycontains(A,B,C,d,e,f).Thiscanleadtoavarietyoffatalproblems.
Cardinality.Cardinalityindicatesthenumberofuniquevaluesfoundwithinadataset.Foraprimarykey,thecardinalityisexpectedtobeequaltothetotalnumberofrecords,whileforaYes/Nofieldthecardinalityisexpectedtobetwo.
Uniqueness.Thedegreeofuniquenessinthedatacanpointtodataqualityproblems.Afieldthatis98%uniquemayindicatebarbageinaprimarykeyfield.
Constants.Constantsindicatethatthesamevalueispresentineveryrecord.Applicationstendtooverlookconstants(sincetheyknowwhat'
ssupposedtobethere),thuscreatingintegrityproblemsindownstreamactivities.Achangeinaconstantusuallyindicatesachangeinprogramlogicfromanupstreamdataproducer.
Outliers.Somedatamayhavecompletelyunexpectedvalues,suchasn
- 配套讲稿:
如PPT文件的首页显示word图标,表示该PPT已包含配套word讲稿。双击word图标可打开word文档。
- 特殊限制:
部分文档作品中含有的国旗、国徽等图片,仅作为作品整体效果示例展示,禁止商用。设计者仅对作品中独创性部分享有著作权。
- 关 键 词:
- Continuous Data Quality Management