大数据技术在建设工程管理中的应用

HOWBIGDATAMAKESCONSTRUCTIONPROJECTRISKINTACT15November20144thICUMASBeijing15November20144thICUMASBeijingRundown(20minutes)WhatisBigDataPitchingonBigDatainHongKongExperienceinaHKlocaluniversityHabitofBigDatapractitionersSomeinterpretationonBigDataEpistemologyValueofBigDataRiskManagementatlargeCorrelationforriskidentificationConstructionHolisticRiskManagement15November20144thICUMASBeijingManagementconceptInfrastructureconceptWhatisBigData?15November20144thICUMASBeijingScatterDiagramWhatisBigData?15November20144thICUMASBeijingBoxPlotWhatisBigData?15November20144thICUMASBeijingBuyausedcar:Step1-GettherawdataYearofmanufacturingOdometerreadingPriceStep2-CleaningupthedataSomeentrieshavewronginformation,someentriesareincomplete.Thereare2002modelslistedfor10lakhs,suchentriesarenotgenuineentries.SoIsortedthedata,foundallsuchentriesanddeletedthem.NowIamleftwithanexcelsheetofaround1000listings.Step3-VisualizingandanalysingWhatisBigData?15November20144thICUMASBeijingWhatisBigData?15November20144thICUMASBeijingWhatisBigData?15November20144thICUMASBeijingStep4-DrawingInferencesSincethepriceofthecardepreciatessharplyduringitsinitialyearsitbettertofindacarthatisaround5yearsold.InthecaseofHondacitythebestdealwouldbecar2007or2008modelthathasnotdonemorethan75,000KMs.Youcandrivethatcarforanother5yearsand75,000KMswithoutlosingalotofitsvalueStatisticalandBayesianinterpretationWhatisBigData?WhatisBigData?15November20144thICUMASBeijing15November20144thICUMASBeijingWhatisBigData?BigDataandepistemologyLucianoFloridisuggestthatscientistsbealerttotheopportunityofdetectingthe‘patterns’(and‘absenceofpatterns’)inthedataasawaytoadvanceknowledge(Floridi,2012).Hewrites“…thepressure…on…geneticsormedicine,experimentalphysicsorneuroscience,istobeabletospotwherethenewpatternswithrealaddedvaluelieintheirimmensedatabasesandhowtheycanbestbeexploitedfor…theadvancementofknowledge.(Floridi,2012)InPlato'sterminologyepisteme(Fricke,2014)meansknowledge,asin"justifiedtruebelief",incontrasttodoxa,commonbelieforopinion.Thewordepistemology,meaningthestudyofknowledge,isderivedfromepisteme.Floridi,L.(2012).BigDataandtheirepistemologicalchallenge.PhilosophyofTechnology,25(4),435-437.Frické,Martin(2014)."Bigdataanditsepistemology." JournaloftheAssociationforInformationScienceandTechnology .15November20144thICUMASBeijingWhatisBigData?15November20144thICUMASBeijingReductionism–anewroletoBigDataWhatisBigData?15November20144thICUMASBeijingValueofBigData找到不变(knowledge)随机应变(agility)WhatisBigData?15November20144thICUMASBeijingRiskmanagementpracticesaremainlyconcerning:•Perceptionsofthetypicaltowardsconstructionriskallocation,andtheimportanceofdifferentriskcategories[13];•Usageoftechniquesatdifferentriskmanagementstages[2];•Usageofriskmanagementtechniquesandbarrierstoriskmanagementinengineeringconstructionindustry[16];•Generalcontractors’perceptiononrisksandtheuseofriskmanagementtechniques[1];•Contractors’applicationofvariousanalyticaltechniquesforriskassessment[18];•VariousrisksperceivedbythecontractorsinChineseconstructionmarket[9];•CriticalrisksassociatedwithChina’sbuild-operate-transferBOTprojectsandtheeffectivenessofmitigationmeasures[22];•PerceptionsofriskallocationintheconstructionindustryofmainlandChinaandHongKong[17];•AllocationofriskinPPP/PFIconstructionprojectsintheUnitedKingdom[4];•Contractualriskandliabilitysharinginhydropowerconstruction[6];•Practicesofusingriskmanagementapproachesinselectedindustries[20];•Riskmanagementservices,tools,andtechniquescurrentlyusedbyconsultants[23];•Riskmanagementintheconceptualphaseofaproject[21].RiskManagementatlargeDecisionmakingCorrelationforriskidentificationImportanceofRiskMuchtimeinconstructionmanagementisspentfocusingonrisksManypracticesinconstructionaredrivenbyriskBondingrequirementsInsuranceLicensingContractstructureGeneralconditionsPaymentTermsDeliveryMethodSelectionmechanismCorrelationforriskidentificationProjectRiskManagementFEASIBILITYDESIGNPLANNINGCLOSEOUTDEVELOPMENTOPERATIONSCorrelationforriskidentificationFinancing&EvaluationRiskAnalysis&AttitudeRiskManagementPhaseFEASIBILITYDESIGNPLANNINGCLOSEOUTDEVELOPMENTOPERATIONSRISKMgtRiskmanagementAssessment,trackingandcontrolTools:RiskHierarchicalmodeling:RiskbreakdownstructuresRiskmatrixesContingencyplan:preventivemeasures,correctiveactions,riskbudget,etc.CorrelationforriskidentificationRiskAssessmentSchemeforqualitativeandquantitativeassessmentsAtallsteps,riskreducingmeasuresneedtobeconsideredCorrelationforriskidentificationSystemdefinitionHazardidentificationAnalysisofaccidentscenariosConsequenceanalysisandmodellingEstimationofaccidentfrequenciesRiskestimationRiskAnalysis–MainStepsCorrelationforriskidentificationRiskAnalysisHazardIdentificationHazard&ScenarioAnalysisLikelihoodConsequencesRisk”Whatif”ChecklistsHAZOPTaskanalysisIndex(Dow,Mond)RiskAnalysis–MainStepsCorrelationforriskidentificationRiskAnalysisHazardIdentificationHazard&ScenarioAnalysisLikelihoodConsequencesRiskFaulttreeanalysisEventtreeanalysisBowtiesBarrierdiagramsReliabilitydataHumanreliabilityConsequencemodelsRiskAnalysis–MainStepsCorrelationforriskidentificationRiskAnalysisHazardIdentificationHazard&ScenarioAnalysisLikelihoodConsequencesRiskIdentifySafetyBarriersSomeRisksWeatherchangesDifferentproductivity(Sub)contractorsareUnreliableLackcapacitytodoworkLackavailabilitytodoworkUnscrupulousFinanciallyunstableLatematerialsdeliveryLawsuitsLabordifficultiesUnexpectedmanufacturingcostsFailuretofindsufficienttenantsCommunityoppositionInfighting&acrimoniousrelationshipsUnrealisticallylowbidLate-stagedesignchangesUnexpectedsubsurfaceconditionsSoiltypeGroundwaterUnexpectedObstaclesSettlementofadjacentstructuresHighlifecyclecostsPermittingproblems…CorrelationforriskidentificationAnalysisUsingDecisionTreesDecisiontreesareapowerfulanalysistoolExampleanalytictechniquesStrategyselection(MonteCarlosimulation)One-wayandmulti-waysensitivityanalysesValueofinformationCorrelationforriskidentificationExample-CompetingBidTreeCorrelationforriskidentificationMonteCarlosimulationMonteCarlosimulationrandomlygeneratesvaluesforuncertainvariablesoverandovertosimulateamodel.It'susedwiththevariablesthathaveaknownrangeofvaluesbutanuncertainvalueforanyparticulartimeorevent.Foreachuncertainvariable,youdefinethepossiblevalueswithaprobabilitydistribution.Distributiontypesinclude:AsimulationcalculatesmultiplescenariosofamodelbyrepeatedlysamplingvaluesfromtheprobabilitydistributionsComputersoftwaretoolscanperformasmanytrials(orscenarios)asyouwantandallowtoselecttheoptimalstrategy…CorrelationforriskidentificationSensitivityAnalysisCorrelationforriskidentification15November20144thICUMASBeijingConstructionHolisticRiskManagementQuestionsAllriskitemsaredeterministicinnature?Heuristic?Howaboutrisksthatwedonotknow?AnycaseofButterflyeffect?“Whatif”thatwedonotknow!15November20144thICUMASBeijing•Read:Initialprocessing,performsentityandfactextraction.SuspiciousActivityReports(SARs)areprocessedandkeyinformation(financialtransactions,accountnumbers,accountowners,addresses,financialtransactions,etc.)isautomaticallyidentifiedbybigdata’smachinelearningroutines;.•Resolve:Assemble,organizeandrelateinformation.Resolveentities,performscategorizationandidentifiessimilarities.KeyinformationwithintheSARisresolvedandcorrelatedwithdatacontainedinaglobalknowledgebaseprovidingaholisticviewofnetworkandrelationshipconnectivity.•Reason:Uncover,compareandcorrelateinformation,performtemporalandgeospatialreasoningandproduceaprivateknowledgegraph.Linkagestoriskitemsareuncoveredandsurfacedinaneasy-to-understandformat.ConstructionHolisticRiskManagement15November20144thICUMASBeijingCo-occurrencenetworksaregenerallyusedtoprovideagraphicvisualizationofpotentialrelationshipsbetweenpeople,organizations,conceptsorotherentitiesrepresentedwithinwrittenmaterial.Thegenerationandvisualizationofco-occurrencenetworkshasbecomepracticalwiththeadventofelectronicallystoredtextamenabletotextmining(contractandclaimsreports).ConstructionHolisticRiskManagement15November20144thICUMASBeijingDimensionalAnalysis&Similarity•Dimensionalanalysisisveryusefulforplanning,presentation,andinterpretationofexperimentaldata.•Dimensionalanalysisisamethodforreducingthenumberandcomplexityofexperimentalvariablesvariablesthataffectagivenphysicalphenomena.ConstructionHolisticRiskManagement19November2014LinkpredictionConstructionHolisticRiskManagement19November2014CasualmodelingConstructionHolisticRiskManagement19November2014ConstructionHolisticRiskManagement19November2014ConstructionHolisticRiskManagementREFERENCES [1] Akintoye,A.S.,andMacleod,M.J.(1997)Riskanalysisandmanagementinconstruction.InternationalJournalofProjectManagement,15(1),31–38.[2] Baker,S.(1999).Riskresponsetechniquesemployedcurrentlyformajorprojects.ConstructionManagementEconomics,17,205–213[3] Belkin,M.,andNiyogi,P.(2003)Laplacianeigenmapsfordimensionalityreductionanddatarepresentation.Neuralcomputation,15(6),1373-1396.[4] Bing,L.,Akintoye,A.,Edwards,P.J.,andHardcastle,C.(2005)TheallocationofriskinPPP/PFIconstructionprojectsintheUK.”InternationalJournalofProjectManagement,23,25–35.[5] Bollier,D.,andFirestone,C.M.(2010)Thepromiseandperilofbigdata:AspenInstitute,CommunicationsandSocietyProgramWashington,DC,USA.[6] Charoenngam,C.,andYeh,C.(1999)Contractualriskandliabilitysharinginhydropowerconstruction.InternationalJournalofProjectManagement,17(1),29–37.[7] Chi,C.F.,Chang,T.C.andHung,K.H.(2004)Significantindustry-sourceofinjuries-accidenttypeforoccupationalfatalitiesinTaiwan.InternationalJournalofIndustrialErgonomics34,77–91.[8] Donoho,D.L.(2000).High-dimensionaldataanalysis:Thecursesandblessingsofdimensionality.AMSMathChallengesLecture,1-32.[9] Fang,D.,Li,M.,Fong,P.S.,andShen,L.(2004)RisksinChineseconstructionmarket—Contractors’perspective.JournalofConstructionEngineeringManagement,130(6),853–861.[10] FerreiraCordeiro,R.L.,TrainaJunior,C.,MachadoTraina,A.J.,López,J.,Kang,U.,&Faloutsos,C.(2011)Clusteringverylargemulti-dimensionaldatasetswithMapReduce.PaperpresentedattheProceedingsofthe17thACMSIGKDDinternationalconferenceonKnowledgediscoveryanddatamining.[11] Gavrishchaka,V.V.,andGanguli,S.B.(2003)Volatilityforecastingfrommultiscaleandhigh-dimensionalmarketdata.Neurocomputing,55(1),285-305.[12] Harper,R.S.andKoehn,E.(1998)ManagingindustrialconstructionsafetyinSoutheastTexas.JournalofConstructionEngineeringandManagement124,452–457.[13] Kangari,R.(1995).RiskmanagementperceptionsandtrendsofU.S.ConstructionJournalofConstructionEngineeringManagement,121(4),422–429[14] Kines,P.(2002)Constructionworkers’fallsthroughroofs:fatalversusseriousinjuries.JournalofSafetyResearch,33,195–208.[15] Larsson,T.J.andField,B.(2002)ThedistributionofoccupationalinjuriesrisksintheVictorianconstructionindustry.SafetyScience40,439–456.15November20144thICUMASBeijingREFERENCES [16] Lyons,T.andSkitmore,M.(2003)ProjectriskmanagementintheQueenslandengineeringconstructionindustry:Asurvey.InternationalJournalofProjectManagement,22,51–61.[17] Rahman,M.M.,andKumaraswamy,M.M.(2002)Jointriskmanagementthroughtransactionallyefficientrelationalcontracting.ConstructionManagementEconomics,20,45–54.[18] Shen,L.Y.(1997)ProjectriskmanagementinHongKong.InternationalJournalofProjectManagement,15(2),pp.101–105.[19] Tsay,Y.J.andChiang,J.Y.(2005)CBAR:anefficientmethodforminingassociationrules.Knowledge-BasedSystems18,99–105.[20] Tummala,V.M.R.,Leung,H.M.,Mok,C.K.,Burchett,J.F.,andLeung,Y.H.(1997)Practices,barriersandbenefitsofusingriskmanagementapproachesinselectedHongKongIndustries.InternationalJournalofProjectManagement,15(5),297–312.[21] Uher,T.E.,andToakley,A.R.(1999)Riskmanagementintheconceptualphaseofaproject.InternationalJournalofProjectManagement,19(3),161–169.[22] Wang,S.Q.,Tiong,L.K.,Ting,S.K.,andAshley,D.(1999)Politicalrisks:AnalysisofkeycontractclausesinChina’sBOTproject.JournalofConstructionEngineeringManagement,125(3),190–197.[23] Wood,G.D.,andEllis,R.C.T.(2003)RiskmanagementpracticesofleadingUKcostconsults.Engineering,Construction,ArchitectureManagement,10(4),254–262[24] Wu,X.,Zhu,X.,Wu,G.-Q.,&Ding,W.(2014).Dataminingwithbigdata.KnowledgeandDataEngineering,IEEETransactions,26(1),97-107.[25] Zhang,C.andZhang,S.(2002).AssociationRuleMining:ModelsandAlgorithms.Springer,NewYork.15November20144thICUMASBeijing