供应链需求预测--rickyblcu

DemandForecastinginaSupplyChain7-1大綱預測在供應鏈的角色預測的特性主要企業預測項目預測的方法與組成時間序列預測預測誤差的衡量指標執行預測的建議CPFR2預測在供應鏈的角色ThebasisforallstrategicandplanningdecisionsinasupplychainExamples:Production:scheduling,inventory,aggregateplanningMarketing:salesforceallocation,promotions,newproductionintroductionFinance:plant/equipmentinvestment,budgetaryplanningPersonnel:workforceplanning,hiring,layoffsAllofthesedecisionsareinterrelated3預測的特性Forecastsarealwayswrong.Shouldincludeexpectedvalueandmeasureoferror.Long-termforecastsarelessaccuratethanshort-termforecasts(forecasthorizonisimportant)Aggregateforecastsaremoreaccuratethandisaggregateforecasts4主要企業預測項目市場需求量母體數預測單位需求量預測驅動變數預測市場佔有率預測企業銷售量預測單價預測生命週期預測5預測的方法主觀法(subjectivemethods)預測人員依個人主觀的判斷進行預測常應用在缺乏歷史資料時透過專家進行主觀預測草根法GrassrootsBottomup市調法MarketresearchLong-rangeNewproductsales歷史類推法Historicalanalogy類似的產品經驗類推DelphiMethod以問卷方式蒐集專家意見以進行預測經由問卷溝通，專家間無直接互動以避免主控性以統計量收斂為停止指標6預測的方法客觀法(objectivemethods)以歷史資料為基礎進行預測TimeSeries(外插法)假設過去之需求資料是未來需求良好指標下，使用歷史資料進行預測，適合當需求環境穩定、無劇烈變動時進行Causal(因果關係法)假設需求與環境中某些因素是高度相關，藉由發現需求與環境因素的相關性去估計未來的需求TransferFunctionModel(轉換函數模式)結合TimeSeries與Causal兩者，經由解釋變數與應變數之歷史資料產生轉換函數，再將解釋變數之預測值代入轉換函數產生應變數之預測值ARIMAT、SARIMAT7需求資料的組成Observeddemand(O)=Systematiccomponent(S)+Randomcomponent(R)Level(currentdeseasonalizeddemand)Trend(growthordeclineindemand)Seasonality(predictableseasonalfluctuation)Systematiccomponent:ExpectedvalueofdemandRandomcomponent:ThepartoftheforecastthatdeviatesfromthesystematiccomponentForecasterror:differencebetweenforecastandactualdemand8需求資料組成的關係類型相乘系統部分＝水準×趨勢×季節性因素相加系統部分＝水準＋趨勢＋季節性因素混合系統部分＝（水準＋趨勢）×季節性因素9時間序列預測Forecastdemandforthenextfourquarters.10時間序列預測11預測的方法StaticAdaptiveMovingaverageSimpleexponentialsmoothingHolt’smodel(withtrend)Winter’smodel(withtrendandseasonality)12預測的 流程 快递问题件怎么处理流程河南自建厂房流程下载关于规范招聘需求审批流程制作流程表下载邮件下载流程设计 UnderstandtheobjectivesofforecastingIntegratedemandplanningandforecastingIdentifymajorfactorsthatinfluencethedemandforecastUnderstandandidentifycustomersegmentsDeterminetheappropriateforecastingtechniqueEstablishperformanceanderrormeasuresfortheforecast13時間序列預測GoalistopredictsystematiccomponentofdemandMultiplicative:(level)(trend)(seasonalfactor)Additive:level+trend+seasonalfactorMixed:(level+trend)(seasonalfactor)StaticmethodsAdaptiveforecasting14靜態法Assumeamixedmodel:Systematiccomponent=(level+trend)(seasonalfactor)Ft+l=[L+(t+l)T]St+l=forecastinperiodtfordemandinperiodt+lL=estimateoflevelforperiod0T=estimateoftrendSt=estimateofseasonalfactorforperiodtDt=actualdemandinperiodtFt=forecastofdemandinperiodt15靜態法EstimatinglevelandtrendEstimatingseasonalfactors16範例資料 分析 定性数据统计分析pdf销售业绩分析模板建筑结构震害分析销售进度分析表京东商城竞争战略分析 產品之需求有季節性的現象每年度之第二季為全年度需求最低之時需求皆是從每年度之第二季遞增至下年度之第一季此需求變化呈現週期現象，每個週期為一年三個週期的需求水準有逐漸上升的趨勢17LevelandTrend因子的估計Beforeestimatinglevelandtrend,demanddatamustbedeseasonalizedDeseasonalizeddemand=demandthatwouldhavebeenobservedintheabsenceofseasonalfluctuationsPeriodicity(p)thenumberofperiodsafterwhichtheseasonalcyclerepeatsitselffordemandatTahoeSalt(Table7.1,Figure7.1)p=418去季節因子的需求資料[Dt-(p/2)+Dt+(p/2)+S2Di]/2pforpevenDt=(sumisfromi=t+1-(p/2)tot+1+(p/2))SDi/pforpodd(sumisfromi=t-(p/2)tot+(p/2)),p/2truncatedtolowerinteger19去季節因子的需求資料Fortheexample,p=4isevenFort=3:D3={D1+D5+Sum(i=2to4)[2Di]}/8={8000+10000+[(2)(13000)+(2)(23000)+(2)(34000)]}/8=19750D4={D2+D6+Sum(i=3to5)[2Di]}/8={13000+18000+[(2)(23000)+(2)(34000)+(2)(10000)]/8=2062520去季節因子的需求資料ThenincludetrendDt=L+tTwhereDt=deseasonalizeddemandinperiodtL=level(deseasonalizeddemandatperiod0)T=trend(rateofgrowthofdeseasonalizeddemand)Trendisdeterminedbylinearregressionusingdeseasonalizeddemandasthedependentvariableandperiodastheindependentvariable(canbedoneinExcel)Intheexample,L=18,439andT=52421需求的時間序列(Figure7.3)22估計季節因子UsethepreviousequationtocalculatedeseasonalizeddemandforeachperiodSt=Dt/Dt=seasonalfactorforperiodtIntheexample,D2=18439+(524)(2)=19487D2=13000S2=13000/19487=0.67Theseasonalfactorsfortheotherperiodsarecalculatedinthesamemanner23估計季節因子(Fig.7.4)24估計季節因子Theoverallseasonalfactorfora“season”isthenobtainedbyaveragingallofthefactorsfora“season”Iftherearerseasonalcycles,forallperiodsoftheformpt+i,1<i<p,theseasonalfactorforseasoniisSi=[Sum(j=0tor-1)Sjp+i]/rIntheexample,thereare3seasonalcyclesinthedataandp=4,soS1=(0.42+0.47+0.52)/3=0.47S2=(0.67+0.83+0.55)/3=0.68S3=(1.15+1.04+1.32)/3=1.17S4=(1.66+1.68+1.66)/3=1.6725預測未來需求Usingtheoriginalequation,wecanforecastthenextfourperiodsofdemand:F13=(L+13T)S1=[18439+(13)(524)](0.47)=11868F14=(L+14T)S2=[18439+(14)(524)](0.68)=17527F15=(L+15T)S3=[18439+(15)(524)](1.17)=30770F16=(L+16T)S4=[18439+(16)(524)](1.67)=4479426動態預測法Theestimatesoflevel,trend,andseasonalityareadjustedaftereachdemandobservationGeneralstepsinadaptiveforecastingMovingaverageSimpleexponentialsmoothingTrend-correctedexponentialsmoothing(Holt’smodel)Trend-andseasonality-correctedexponentialsmoothing(Winter’smodel)27動態預測模式的符號說明Ft+1=(Lt+lT)St+1=forecastforperiodt+linperiodtLt=EstimateoflevelattheendofperiodtTt=EstimateoftrendattheendofperiodtSt=EstimateofseasonalfactorforperiodtFt=Forecastofdemandforperiodt(madeperiodt-1orearlier)Dt=ActualdemandobservedinperiodtEt=ForecasterrorinperiodtAt=Absolutedeviationforperiodt=|Et|MAD=MeanAbsoluteDeviation=averagevalueofAt28動態預測的基本步驟Initialize:Computeinitialestimatesoflevel(L0),trend(T0),andseasonalfactors(S1,…,Sp).Thisisdoneasinstaticforecasting.Forecast:Forecastdemandforperiodt+1usingthegeneralequationEstimateerror:ComputeerrorEt+1=Ft+1-Dt+1Modifyestimates:Modifytheestimatesoflevel(Lt+1),trend(Tt+1),andseasonalfactor(St+p+1),giventheerrorEt+1intheforecastRepeatsteps2,3,and4foreachsubsequentperiod29移動平均法UsedwhendemandhasnoobservabletrendorseasonalitySystematiccomponentofdemand=levelThelevelinperiodtistheaveragedemandoverthelastNperiods(theN-periodmovingaverage)CurrentforecastforallfutureperiodsisthesameandisbasedonthecurrentestimateofthelevelLt=(Dt+Dt-1+…+Dt-N+1)/NFt+1=LtandFt+n=LtAfterobservingthedemandforperiodt+1,revisetheestimatesasfollows:Lt+1=(Dt+1+Dt+…+Dt-N+2)/NFt+2=Lt+130移動平均法FromTahoeSaltexample(Table7.1)Attheendofperiod4,whatistheforecastdemandforperiods5through8usinga4-periodmovingaverage?L4=(D4+D3+D2+D1)/4=(34000+23000+13000+8000)/4=19500F5=19500=F6=F7=F8Observedemandinperiod5tobeD5=10000Forecasterrorinperiod5,E5=F5-D5=19500-10000=9500Reviseestimateoflevelinperiod5:L5=(D5+D4+D3+D2)/4=(10000+34000+23000+13000)/4=20000F6=L5=2000031簡單指數平滑法UsedwhendemandhasnoobservabletrendorseasonalitySystematiccomponentofdemand=levelInitialestimateoflevel,L0,assumedtobetheaverageofallhistoricaldataL0=[Sum(i=1ton)Di]/nCurrentforecastforallfutureperiodsisequaltothecurrentestimateofthelevelandisgivenasfollows:Ft+1=LtandFt+n=LtAfterobservingdemandDt+1,revisetheestimateofthelevel:Lt+1=aDt+1+(1-a)LtLt+1=Sum(n=0tot+1)[a(1-a)nDt+1-n]32簡單指數平滑法FromTahoeSaltdata,forecastdemandforperiod1usingexponentialsmoothingL0=averageofall12periodsofdata=Sum(i=1to12)[Di]/12=22083F1=L0=22083Observeddemandforperiod1=D1=8000Forecasterrorforperiod1,E1,isasfollows:E1=F1-D1=22083-8000=14083Assuminga=0.1,revisedestimateoflevelforperiod1:L1=aD1+(1-a)L0=(0.1)(8000)+(0.9)(22083)=20675F2=L1=20675Notethattheestimateoflevelforperiod1islowerthaninperiod033Holt’sModelAppropriatewhenthedemandisassumedtohavealevelandtrendinthesystematiccomponentofdemandbutnoseasonalityObtaininitialestimateoflevelandtrendbyrunningalinearregressionofthefollowingform:Dt=at+bT0=aL0=bInperiodt,theforecastforfutureperiodsisexpressedasfollows:Ft+1=Lt+TtFt+n=Lt+nTt34Holt’sModelAfterobservingdemandforperiodt,revisetheestimatesforlevelandtrendasfollows:Lt+1=aDt+1+(1-a)(Lt+Tt)Tt+1=b(Lt+1-Lt)+(1-b)Tta=smoothingconstantforlevelb=smoothingconstantfortrendExample:TahoeSaltdemanddata.Forecastdemandforperiod1usingHolt’smodel(trendcorrectedexponentialsmoothing)Usinglinearregression,L0=12015(linearintercept)T0=1549(linearslope)35Holt’sModelForecastforperiod1:F1=L0+T0=12015+1549=13564Observeddemandforperiod1=D1=8000E1=F1-D1=13564-8000=5564Assumea=0.1,b=0.2L1=aD1+(1-a)(L0+T0)=(0.1)(8000)+(0.9)(13564)=13008T1=b(L1-L0)+(1-b)T0=(0.2)(13008-12015)+(0.8)(1549)=1438F2=L1+T1=13008+1438=14446F5=L1+4T1=13008+(4)(1438)=1876036Winter’sModelAppropriatewhenthesystematiccomponentofdemandisassumedtohavealevel,trend,andseasonalfactorSystematiccomponent=(level+trend)(seasonalfactor)AssumeperiodicitypObtaininitialestimatesoflevel(L0),trend(T0),seasonalfactors(S1,…,Sp)usingprocedureforstaticforecastingInperiodt,theforecastforfutureperiodsisgivenby:Ft+1=(Lt+Tt)(St+1)andFt+n=(Lt+nTt)St+n37Winter’sModelAfterobservingdemandforperiodt+1,reviseestimatesforlevel,trend,andseasonalfactorsasfollows:Lt+1=a(Dt+1/St+1)+(1-a)(Lt+Tt)Tt+1=b(Lt+1-Lt)+(1-b)TtSt+p+1=g(Dt+1/Lt+1)+(1-g)St+1a=smoothingconstantforlevelb=smoothingconstantfortrendg=smoothingconstantforseasonalfactor38Winter’sModelExample:TahoeSaltdata.Forecastdemandforperiod1usingWinter’smodel.Initialestimatesoflevel,trend,andseasonalfactorsareobtainedasinthestaticforecastingcase39Winter’sModelL0=18439T0=524S1=0.47,S2=0.68,S3=1.17,S4=1.67F1=(L0+T0)S1=(18439+524)(0.47)=8913Theobserveddemandforperiod1=D1=8000Forecasterrorforperiod1=E1=F1-D1=8913-8000=913Assumea=0.1,b=0.2,g=0.1;reviseestimatesforlevelandtrendforperiod1andforseasonalfactorforperiod5L1=a(D1/S1)+(1-a)(L0+T0)=(0.1)(8000/0.47)+(0.9)(18439+524)=18769T1=b(L1-L0)+(1-b)T0=(0.2)(18769-18439)+(0.8)(524)=485S5=g(D1/L1)+(1-g)S1=(0.1)(8000/18769)+(0.9)(0.47)=0.47F2=(L1+T1)S2=(18769+485)(0.68)=1309340預測誤差類型偏差(bias)與隨機誤差(randomerror)偏差的原因未涵蓋正確之變數錯誤的變數關係錯誤的趨勢線錯誤的季節需求修正未發現的長期趨勢41預測誤差的衡量指標Forecasterror=Et=Ft-DtMeansquarederror(MSE)MSEn=(Sum(t=1ton)[Et2])/nAbsolutedeviation=At=|Et|Meanabsolutedeviation(MAD)MADn=(Sum(t=1ton)[At])/ns=1.25MAD42預測誤差的衡量指標Meanabsolutepercentageerror(MAPE)MAPEn=(Sum(t=1ton)[|Et/Dt|100])/nBiasShowswhethertheforecastconsistentlyunder-oroverestimatesdemand;shouldfluctuatearound0biasn=Sum(t=1ton)[Et]TrackingsignalShouldbewithintherangeof+6Otherwise,possiblyuseanewforecastingmethodTSt=bias/MADt43執行預測的建議CollaborateinbuildingforecastsBesuretodistinguishbetweendemandandsales44協同規劃預測補貨模式(CollaborativePlanning,Forecasting,andReplenishment,CPFR)源起1995年，Wal-Mart、Warner-Lambert、SAP、Manugistics、BenchmarkingPartners等五家公司合作之零售業供應鏈工作小組進行之供應商與零售商快速回應示範計劃共同標準協會(VoluntaryInterindustryCommunicationStandards,VICS)以示範計劃中有關買賣雙方合作預測市場需求內容為基礎，於1998年正式發表有關供應鏈合作商務的規範，以指導供應商與零售商進行協同預測與例外狀況之處理機制演進成為B2B電子商務中供應鏈合作商務規範。45CPFR基本原則零售商與供應商間:建立一致且有效率的需求預測與規劃資訊共享與風險共擔的營運方式交易流程整合 制度 关于办公室下班关闭电源制度矿山事故隐患举报和奖励制度制度下载人事管理制度doc盘点制度下载 化例外狀況處理46CPFR主要內容企業間商務流程定義與規範Nine-StepProcessModelCPFR導入規範RoadmapofCPFR績效評估指標Metrics,KeyPerformanceIndicators標準商業訊息CPFRXMLEAN.UCCCPFRXML479步驟流程模式概觀48演讲完毕，谢谢观看！