版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進行舉報或認領(lǐng)
文檔簡介
SupplyChainManagement:Strategy,Planning,andOperationSeventhEditionChapter7DemandForecastinginaSupplyChainCopyright?2019,2016,2013PearsonEducation,Inc.AllRightsReservedLearningObjectives7.1Understandtheroleofforecastingforbothanenterpriseandasupplychain.7.2Identifythecomponentsofademandforecastandsomebasicapproachestoforecasting.7.3Forecastdemandusingtime-seriesmethodologiesgivenhistoricaldemanddatainasupplychain.7.4Analyzedemandforecaststoestimateforecasterror.7.5UseExceltobuildtime-seriesforecastingmodels.RoleofForecastinginaSupplyChainThebasisforallplanningdecisionsinasupplychainUsedforbothpushandpullprocessesProductionscheduling,inventory,aggregateplanningSalesforceallocation,promotions,newproductionintroductionPlant/equipmentinvestment,budgetaryplanningWorkforceplanning,hiring,layoffsAllofthesedecisionsareinterrelatedCharacteristicsofForecastsForecastsarealwaysinaccurateandshouldthusincludeboththeexpectedvalueoftheforecastandameasureofforecasterrorLong-termforecastsareusuallylessaccuratethanshort-termforecastsAggregateforecastsareusuallymoreaccuratethandisaggregateforecastsIngeneral,thefartherupthesupplychainacompanyis,thegreateristhedistortionofinformationitreceivesSummaryofLearningObjective1(1of2)Forecastingisakeyinputforvirtuallyeverydesignandplanningdecisionmadeinasupplychain.Itisimportanttorecognizethatallforecastsarelikelytobewrong.Thus,anestimationofforecasterrorisessentialtoeffectivelyusetheforecast.Reducingtheforecasthorizon(byreducingtheleadtimeoftheassociateddecision)andaggregationaretwoeffectiveapproachestodecreaseforecasterror.SummaryofLearningObjective1(2of2)Arelativelyrecentphenomenon,however,istocreatecollaborativeforecastsforanentiresupplychainandusetheseasthebasisfordecisions.Collaborativeforecastinggreatlyincreasestheaccuracyofforecastsandallowsthesupplychaintomaximizeitsperformance.Withoutcollaboration,supplychainstagesfartherfromdemandwilllikelyhavepoorforecaststhatwillleadtosupplychaininefficienciesandalackofresponsiveness.ComponentsandMethods(1of2)CompaniesmustidentifythefactorsthatinfluencefuturedemandandthenascertaintherelationshipbetweenthesefactorsandfuturedemandPastdemandLeadtimeofproductreplenishmentPlannedadvertisingormarketingeffortsPlannedpricediscountsStateoftheeconomyActionsthatcompetitorshavetakenComponentsandMethods(2of2)QualitativePrimarilysubjectiveRelyonjudgmentTimeSeriesUsehistoricaldemandonlyBestwithstabledemandCausalRelationshipbetweendemandandsomeotherfactorSimulationImitateconsumerchoicesthatgiverisetodemandComponentsofAnObservationObserveddemand(O)=systematiccomponent(S)
+randomcomponent(R)Systematiccomponent–expectedvalueofdemandLevel(currentdeseasonalizeddemand)Trend(growthordeclineindemand)Seasonality(predictableseasonalfluctuation)Randomcomponent–partofforecastthatdeviatesfromsystematicpartForecasterror–differencebetweenforecastandactualdemandFiveImportantPointsintheForecastingProcessUnderstandtheobjectiveofforecasting.Integratedemandplanningandforecastingthroughoutthesupplychain.Identifythemajorfactorsthatinfluencethedemandforecast.Forecastattheappropriatelevelofaggregation.Establishperformanceanderrormeasuresfortheforecast.SummaryofLearningObjective2Demandconsistsofasystematicandarandomcomponent.Thesystematiccomponentmeasurestheexpectedvalueofdemand.Therandomcomponentmeasuresfluctuationsindemandfromtheexpectedvalue.Thesystematiccomponentconsistsoflevel,trend,andseasonality.Levelmeasuresthecurrentde-seasonalizeddemand.Trendmeasuresthecurrentrateofgrowthordeclineindemand.Seasonalityindicatespredictableseasonalfluctuationsindemand.Thegoalofforecastingistoestimatethesystematiccomponentandthesize(notdirection)oftherandomcomponent(intheformofaforecasterror).Goodforecastingrequiresaclearunderstandingoftheobjectiveoftheforecastandshouldbeintegratedacrossthesupplychain.Time-SeriesForecastingMethodsThreewaystocalculatethesystematiccomponentMultiplicativeS=level×trend×seasonalfactorAdditiveS=level+trend+seasonalfactorMixedS=(level+trend)×seasonalfactorStaticMethodsSystematiccomponent=(level+trend)×seasonalfactorWhereL = estimateoflevelatt=0T = estimateoftrendSt = estimateofseasonalfactorforPeriodtDt = actualdemandobservedinPeriodtFt = forecastofdemandforPeriodtTahoeSalt(1of5)Table7-1QuarterlyDemandforTahoeSaltYearQuarterPeriod,tDemand,Dt121 8,000132 13,000143 23,000214 34,000225 10,000236 18,000247 23,000TahoeSalt(2of5)Table7-1[continued]YearQuarterPeriod,tDemand,Dt318 38,000329 12,0003310 13,0003411 32,0004112 41,000TahoeSalt(3of5)Figure7-1QuarterlyDemandatTahoeSaltDeseasonalizedemandandrunlinearregressiontoestimatelevelandtrend.Estimateseasonalfactors.EstimateLevelandTrend(1of2)Periodicityp=4,t=3EstimateLevelandTrend(2of2)TahoeSalt(4of5)Figure7-2ExcelWorkbookwithDeseasonalizedDemandforTahoeSaltTahoeSalt(5of5)Figure7-3DeseasonalizedDemandforTahoeSaltAlinearrelationshipexistsbetweenthedeseasonalizeddemandandtimebasedonthechangeindemandovertimeEstimatingSeasonalFactors(1of3)Figure7-4DeseasonalizedDemandandSeasonalFactorsforTahoeSaltEstimatingSeasonalFactors(2of3)EstimatingSeasonalFactors(3of3)AdaptiveForecasting(1of2)Theestimatesoflevel,trend,andseasonalityareupdatedaftereachdemandobservationEstimatesincorporateallnewdatathatareobservedAdaptiveForecasting(2of2)WhereLt=estimateoflevelattheendofPeriodt
Tt=estimateoftrendattheendofPeriodt
St=estimateofseasonalfactorforPeriodt
Ft
=forecastofdemandforPeriodt(madePeriodt–1orearlier)Dt=actualdemandobservedinPeriodt
Et=Ft
–Dt=forecasterrorinPeriodtStepsinAdaptiveForecastingInitializeComputeinitialestimatesoflevel(L0),trend(T0),andseasonalfactors(S1,…,Sp)ForecastForecastdemandforperiodt+1EstimateerrorComputeerrorEt+1=Ft+1–Dt+1ModifyestimatesModifytheestimatesoflevel(Lt+1),trend(Tt+1),andseasonalfactor(St+p+1),giventheerrorEt+1MovingAverageUsedwhendemandhasnoobservabletrendorseasonalitySystematiccomponentofdemand=levelThelevelinperiodtistheaveragedemandoverthelastN
periodsAfterobservingthedemandforperiodt
+1,revisetheestimatesMovingAverageExample(1of2)AsupermarkethasexperiencedweeklydemandofmilkofD1=120,D2=127,D3=114,andD4=122gallonsoverthepastfourweeksForecastdemandforPeriod5usingafour-periodmovingaverageWhatistheforecasterrorifdemandinPeriod5turnsouttobe125gallons?MovingAverageExample(2of2)ForecastdemandforPeriod5 F5=L4=120.75gallonsErrorifdemandinPeriod5=125gallons E5=F5–D5=120.75–125=–4.25ReviseddemandSimpleExponentialSmoothing(1of3)UsedwhendemandhasnoobservabletrendorseasonalitySystematiccomponentofdemand=levelInitialestimateoflevel,L0,assumedtobetheaverageofallhistoricaldataSimpleExponentialSmoothing(2of3)GivendataforPeriods1tonCurrentforecastRevisedforecastusingsmoothingconstant(0<α<1)ThusSimpleExponentialSmoothing(3of3)SupermarketdataF1
=L0
=120.75E1
=F1?D1
=120.75?120=0.75L1=αD1+(1?α)L0=0.1×120+0.9×120.75=120.68Trend-CorrectedExponentialSmoothing(Holt’sModel)(1of4)AppropriatewhenthedemandisassumedtohavealevelandtrendinthesystematiccomponentofdemandbutnoseasonalitySystematiccomponentofdemand=level+trendTrend-CorrectedExponentialSmoothing(Holt’sModel)(2of4)ObtaininitialestimateoflevelandtrendbyrunningalinearregressionDt
=at+bT0=a,L0=bInPeriodt,theforecastforfutureperiodsisFt+1
=Lt+Tt
andFt+n=Lt+nTtRevisedestimatesforPeriodtTrend-CorrectedExponentialSmoothing(Holt’sModel)(3of4)SmartphoneplayerdemandD1=8,415,D2=8,732,D3=9,014,D4=9,808,D5=10,413,D6=11,961,α=0.1,β=0.2UsingregressionanalysisL0=7,367andT0=673ForecastforPeriod1F1=L0+T0=7,367+673=8,040Period1errorE1=F1–D1=8,040–8,415=–375Trend-CorrectedExponentialSmoothing(Holt’sModel)(4of4)RevisedestimateWithnewL1F2=L1+T1=8,078+681=8,759ContinuingF7=L6+T6=11,399+673=12,072Trend-andSeasonality-CorrectedExponentialSmoothing(1of2)Appropriatewhenthesystematiccomponentofdemandhasalevel,trend,andseasonalfactorSystematiccomponent=(level+trend)×seasonalfactorTrend-andSeasonality-CorrectedExponentialSmoothing(2of2)Afterobservingdemandforperiodt+1,reviseestimatesforlevel,trend,andseasonalfactorsα
=smoothingconstantforlevelβ
=smoothingconstantfortrendγ
=smoothingconstantforseasonalfactorWinter’sModel(1of3)L0=18,439T0=524S1=0.47,S2=0.68,S3=1.17,S4=1.67F1=(L0+T0)S1=(18,439+524)(0.47)=8,913TheobserveddemandforPeriod1=D1=8,000ForecasterrorforPeriod1=E1=F1–D1=8,913–8,000=913Winter’sModel(2of3)Assumeα
=0.1,β
=0.2,γ
=0.1;reviseestimatesforlevelandtrendforperiod1andforseasonalfactorforPeriod5Winter’sModel(3of3)ForecastdemandforPeriod2F2
=(L1+T1)S2=(18,769+485)(0.68)=13,093TimeSeriesModelsForecastingMethodApplicabilityMovingaverageNotrendorseasonalitySimpleexponentialsmoothingNotrendorseasonalityHolt’smodelTrendbutnoseasonalityWinter’smodelTrendandseasonalitySummaryofLearningObjective3Time-seriesmethodsforforecastingarecategorizedasstaticoradaptive.Instaticmethods,theestimatesofparametersarenotupdatedasnewdemandisobserved.Staticmethodsincluderegression.Inadaptivemethods,theestimatesareupdatedeachtimeanewdemandisobserved.Adaptivemethodsincludemovingaverages,simpleexponentialsmoothing,Holt’smodel,andWinter’smodel.Movingaveragesandsimpleexponentialsmoothingarebestusedwhendemanddisplaysneithertrendnorseasonality.Holt’smodelisbestwhendemanddisplaysatrendbutnoseasonality.Winter’smodelisappropriatewhendemanddisplaysbothtrendandseasonality.MeasuresofForecastError(1of2)Forecasterrorscontainvaluableinformationandmustbeanalyzedfortworeasons:ManagersuseerroranalysistodeterminewhetherthecurrentforecastingmethodispredictingthesystematiccomponentofdemandaccuratelyAllcontingencyplansmustaccountforforecasterrorMeasuresofForecastError(2of2)SummaryofLearningObjective4Forecasterrormeasurestherandomcomponentofdemand.Thismeasureisimportantbecauseitrevealshowinaccurateaforecastislikelytobeandwhatcontingenciesafirmmayhavetoplanfor.TheM
S
E,M
A
D,andM
A
P
Eareusedtoestimatethesizeofthefore-casterror.ThebiasandT
Sareusedtoestimateiftheforecastconsistentlyover-orunder-forecastsorifdemandhasdeviatedsignificantlyfromhistoricalnorms.SelectingtheBestSmoothingConstant(1of2)Figure7-5SelectingSmoothingConstantbyMinimizingM
S
ESelectingtheBestSmoothingConstant(2of2)Figure7-6SelectingSmoothingConstantbyMinimizingM
A
DForecastingDemandatTahoeSalt(1of10)MovingaverageSimpleexponentialsmoothingTrend-correctedexponentialsmoothingTrend-andseasonality-correctedexponentialsmoothingForecastingDemandatTahoeSalt(2of10)Figure7-7TahoeSaltForecastsUsingFour-PeriodMovingAverageForecastingDemandatTahoeSalt(3of10)MovingaverageL12=24,500F13=F14=F15=F16=L12=24,500σ
=1.25×9,719=12,148ForecastingDemandatTahoeSalt(4of10)Figure7-8TahoeSaltForecastsUsingSimpleExponentialSmoothingForecastingDemandatTahoeSalt(5of10)Simpleexponentialsmoothingα
=0.1L0=22,083L12=23,490F13=F14=F15=F16=L12=23,490σ=1.25×
10,208=12,761ForecastingDemandatTahoeSalt(6of10)Figure7-9Trend-CorrectedExponentialSmoothingForecastingDemandatTahoeSalt(7of10)Trend-CorrectedExponentialSmoothingL0=12,015andT0=1,549L12=30,443andT12=1,541F13=L12+T12=30,443+1,541=31,984F14=L12+2T12=30,443+2×1,541=33,525F15=L12+3T12=30,443+3×
1,541=35,066F16=L12+4T12=30,443+4×
1,541=36,607σ
=1.25×
8,836=11,045ForecastingDemandatTahoeSalt(8of10)Figure7-10Trend
溫馨提示
- 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
- 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
- 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
- 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
- 5. 人人文庫網(wǎng)僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負責。
- 6. 下載文件中如有侵權(quán)或不適當內(nèi)容,請與我們聯(lián)系,我們立即糾正。
- 7. 本站不保證下載資源的準確性、安全性和完整性, 同時也不承擔用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。
最新文檔
- 二零二五年度全新店面轉(zhuǎn)讓定金及風險管理協(xié)議3篇
- 2025年度5G通信技術(shù)應(yīng)用合作協(xié)議范例3篇
- 2025年度內(nèi)墻膩子施工與廢棄物處理技術(shù)合作勞務(wù)合同2篇
- 2025年度旅游項目承包合同2篇
- 2025年度文化產(chǎn)業(yè)資產(chǎn)并購收購協(xié)議書3篇
- 2025年度內(nèi)部承包合同協(xié)議書:XX工廠內(nèi)部承包生產(chǎn)任務(wù)分配與考核協(xié)議3篇
- 2025汽車租賃合同樣本范文
- 2025年度跨境電商全新員工入職與全球業(yè)務(wù)拓展合同3篇
- 2025年度公司車輛租賃及駕駛員培訓(xùn)考核合同3篇
- 二零二五年度智慧教育平臺合作項目協(xié)議書模板3篇
- 2024-2030年中國高密度聚乙烯管道行業(yè)發(fā)展展望與投資策略建議報告
- 2024-2030年中國醋酸乙烯行業(yè)運營狀況與發(fā)展風險評估報告
- 企業(yè)文化塑造與員工激勵方案
- 2023-2024學年貴州省遵義市新蒲新區(qū)八年級(上)期末數(shù)學試卷(含答案)
- 2022屆河北省石家莊市高一上學期期末考試化學試題(含解析)
- 2025年日歷臺歷中文版縱向排版帶節(jié)假日調(diào)休周日開始
- 25題電控工程師崗位常見面試問題含HR問題考察點及參考回答
- 小學一年級班會課教案匯編 全冊
- 公司董事會、總經(jīng)理辦公會議事清單.docx
- 煤礦礦井供電設(shè)計(DOC26頁)
- 中國鶴翔莊氣功之五站樁功
評論
0/150
提交評論