p2oc804OLAP.pptVIP免费

11DataMining:ConceptsandTechniques(3rded.)—Chapter4—JiaweiHan,MichelineKamber,andJianPeiUniversityofIllinoisatUrbana-Champaign&SimonFraserUniversity©2011Han,Kamber&Pei.Allrightsreserved.3Chapter4:DataWarehousingandOn-lineAnalyticalProcessingDataWarehouse:BasicConceptsDataWarehouseModeling:DataCubeandOLAPDataWarehouseDesignandUsageDataWarehouseImplementationDataGeneralizationbyAttribute-OrientedInductionSummary4WhatisaDataWarehouse?Definedinmanydifferentways,butnotrigorously.Adecisionsupportdatabasethatismaintainedseparatelyfromtheorganization’soperationaldatabaseSupportinformationprocessingbyprovidingasolidplatformofconsolidated,historicaldataforanalysis.“Adatawarehouseisasubject-oriented,integrated,time-variant,andnonvolatilecollectionofdatainsupportofmanagement’sdecision-makingprocess.”—W.H.InmonDatawarehousing:Theprocessofconstructingandusingdatawarehouses5DataWarehouse—Subject-OrientedOrganizedaroundmajorsubjects,suchascustomer,product,salesFocusingonthemodelingandanalysisofdatafordecisionmakers,notondailyoperationsortransactionprocessingProvideasimpleandconciseviewaroundparticularsubjectissuesbyexcludingdatathatarenotusefulinthedecisionsupportprocess6DataWarehouse—IntegratedConstructedbyintegratingmultiple,heterogeneousdatasourcesrelationaldatabases,flatfiles,on-linetransactionrecordsDatacleaninganddataintegrationtechniquesareapplied.Ensureconsistencyinnamingconventions,encodingstructures,attributemeasures,etc.amongdifferentdatasourcesE.g.,Hotelprice:currency,tax,breakfastcovered,etc.Whendataismovedtothewarehouse,itisconverted.7DataWarehouse—TimeVariantThetimehorizonforthedatawarehouseissignificantlylongerthanthatofoperationalsystemsOperationaldatabase:currentvaluedataDatawarehousedata:provideinformationfromahistoricalperspective(e.g.,past5-10years)EverykeystructureinthedatawarehouseContainsanelementoftime,explicitlyorimplicitlyButthekeyofoperationaldatamayormaynotcontain“timeelement”8DataWar...