操作管理-决策工具(2)

1998byPrenticeHall,Inc.2OperationsManagementDecision-MakingToolsModuleA1998byPrenticeHall,Inc.3Outline•TheDecisionProcessinOperations•FundamentalsofDecisionMaking•DecisionTables–DecisionMakingUnderRisk–ExpectedValueofPerfectInformation(EVPI)•DecisionTrees•AMoreComplexDecisionTree1998byPrenticeHall,Inc.4LearningObjectives•Statethedecisionprocess•Describedecisiontables&trees•Explaindecisionmakingunderrisk•Computeexpectedmonetaryvalue•Usedecisiontables&treestomakedecisions1998byPrenticeHall,Inc.5Models,andtheTechniquesofScientificManagementCanHelpManagersTo:•Gaindeeperinsightintothenatureofbusinessrelationships•Findbetterwaystoassessvaluesinsuchrelationships;and•Seeawayofreducing,oratleastunderstanding,uncertaintythatsurroundsbusinessplansandactions1998byPrenticeHall,Inc.6StepstoGoodDecisions•Defineproblem&influencingfactors•Establishdecisioncriteria•Selectdecision-makingtool(model)•Identify&evaluatealternativesusingdecision-makingtool(model)•Selectbestalternative•Implementdecision1998byPrenticeHall,Inc.7AdvantagesofUsingModels•Theyarelessexpensiveanddisruptivethanexperimentingwiththerealworldsystem•Theyallowoperationsmanagerstoask“Whatif”typesofquestions•Theyarebuiltformanagementproblemsandencouragemanagementinput•Theyforceaconsistentandsystematicapproachtotheanalysisofproblems•Theyrequiremanagerstobespecificaboutconstraintsandgoalsrelatingtoaproblem•Theycanhelpreducethetimeneededindecisionmaking1998byPrenticeHall,Inc.8LimitationsofModels•Theymaybeexpensiveandtime-consumingtodevelopandtest•Theyareoftenmisusedandmisunderstood(andfeared)becauseoftheirmathematicalcomplexity•Theytendtodownplaytheroleandvalueofnonquantifiableinformation•Theyoftenhaveassumptionsthatoversimplifythevariablesoftherealworld1998byPrenticeHall,Inc.9决策过程TheDecision-MakingProcess问题Problem决策Decision计量分析QuantitativeAnaly逻辑Logic历史资料HistoricalData行销研究MarketingResearch科学分析ScientificAnalysis模式Modeling计质分析QualitativeAnalysis感情Emotions直观Intuition个人经验和激励PersonalExperienceandMotivation谣言Rumors1998byPrenticeHall,Inc.10决策问题DecisionProblem方案Alternatives自然的状况StatesofNature结果Out-comes问题的元素ElementsofaProblem1998byPrenticeHall,Inc.11DecisionProblemAlternativesStatesofNatureOut-comes•决策树Decisiontrees•决策表Decisiontables呈现决策问题解决的方法1998byPrenticeHall,Inc.12决策论FundamentalsofDecisionTheory有三种决策模式:•在确定环境下作决策Decisionmakingundercertainty•在风险下作决策Decisionmakingunderrisk•在不确定环境下作决策Decisionmakingunderuncertainty条件Terms:•方案Alternative:courseofactionorchoice•状况Stateofnature:anoccurrenceoverwhichthedecisionmakerhasnocontrol决策树的符号Symbolsusedindecisiontree:决策点Adecisionnodefromwhichoneofseveralalternativesmaybeselected决策状况Astateofnaturenodeoutofwhichonestateofnaturewilloccur1998byPrenticeHall,Inc.13状况StatesofNature方案AlternativesState1State2Alternative1Outcome1Outcome2Alternative2Outcome3Outcome4决策表DecisionTable替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216报酬表确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)2161998byPrenticeHall,Inc.16•概率性决策Probabilisticdecisionsituation•不同状况的概率性决策Statesofnaturehaveprobabilitiesofoccurrence•选择最大的金钱期望回报Selectalternativewithlargestexpectedmonetaryvalue(EMV)–Averagereturnforalternative•Ifdecisionwererepeatedmanytimes在有风险下进行决策DecisionMakingUnderRisk不确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216小中取大(10)大中取大平均值或Laplace最大遗憾中取最小不确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216小中取大(10)大中取大(16)平均值或Laplace最大遗憾中取最小不确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216小中取大(10)大中取大(16)平均值或Laplace(10.33)最大遗憾中取最小平均数平均数平均数最大遗憾值取最小不确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216小中取大(10)大中取大(16)平均值或Laplace(10.33)最大遗憾中取最小10-10,10-7,10+412-10,12-12,12-216-1016-1216-16不确定环境下的决策替代方案低中高小型设备$10$10$10中型设备$7$12$12大型设备(4)216小中取大(10)大中取大(16)平均值或Laplace(10.33)最大遗憾中取最小(4)最大遗憾值取最小026304141001998byPrenticeHall,Inc.22EMVAVPVVPVVPVVPViiiiNN(()()()())=N=S*=*+*++*11122LNo.ofstatesofnaturePayoffProbabilityofpayoffAlternativeiEMV方程式ExpectedMonetaryValueEquation1998byPrenticeHall,Inc.23•以图解进行决策Graphicaldisplayofdecisionprocess•问题解决Usedforsolvingproblems–With1setofalternatives&statesofnature•Decisiontablescanbeusedalso–Withseveralsetsofalternatives&statesofnature(sequentialdecisions)•Decisiontablescannotbeused•EMViscriterionusedmostoften决策树DecisionTrees1998byPrenticeHall,Inc.24应用决策树来解决问题AnalyzingProblemswithDecisionTrees•问题的定义Definetheproblem•绘决策树Structureordrawthedecisiontree•决定不同的决策概率Assignprobabilitiestothestatesofnature•计算个别状况的回报Estimatepayoffsforeachpossiblecombinationofalternativesandstatesofnature•计算它的投资回报率Solvetheproblembycomputingexpectedmonetaryvaluesforeachstate-of-naturenode1998byPrenticeHall,Inc.2512State1State2State1State2DecisionNodeOutcome1Outcome2Outcome3Outcome4StateofNatureNodeDecisionTree决策树替代方案低(.3)中(.5)高(.2)小型设备$10$10$10中型设备$7$12$12大型设备(4)216=0.3x10+10x0.5+10x0.2=100.3x7+0.5x12+12x0.2=10.5-0.3x4+2x0.5+16x0.2=3EV/PI大型中型小型低低低中中中高高高-4216712121010100.30.30.30.50.50.50.20.20.2EVPI=0.3x10+0.5x12+0.2x16-10.5=12.2-10.5=1.7310.510