Bài giảng Operations Management - Chapter 4: Forecasting

Operations ManagementForecastingChapter 41OutlineGLOBAL COMPANY PROFILE: TUPPERWARE CORPORATIONWHAT IS FORECASTING?Forecasting Time HorizonsThe Influence of Product Life CycleTYPES OF FORECASTSTHE STRATEGIC IMPORTANCE OF FORECASTINGHuman ResourcesCapacitySupply-Chain ManagementSEVEN STEPS IN THE FORECASTING SYSTEM2© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Outline - ContinuedFORECASTING APPROACHESOverview of Qualitative MethodsOverview of Quantitative MethodsTIME-SERIES FORECASTINGDecomposition of Time SeriesNaïve ApproachMoving AveragesExponential SmoothingExponential Smoothing with Trend AdjustmentTrend ProjectionsSeasonal Variations in DataCyclic Variations in Data3© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Outline - ContinuedASSOCIATIVE FORECASTING METHODS: REGRESSION AND CORRELATION ANALYSISUsing Regression Analysis to ForecastStandard Error of the EstimateCorrelation Coefficients for Regression LinesMultiple-Regression AnalysisMONITORING AND CONTROLLING FORECASTSAdaptive SmoothingFocus ForecastingFORECASTING IN THE SERVICE SECTOR4© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Learning ObjectivesWhen you complete this chapter, you should be able to :Identify or Define:ForecastingTypes of forecastsTime horizonsApproaches to forecasts5© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Learning Objectives - continuedWhen you complete this chapter, you should be able to :Describe or Explain:Moving averagesExponential smoothingTrend projectionsRegression and correlation analysisMeasures of forecast accuracy6© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Forecasting at TupperwareEach of 50 profit centers around the world is responsible for computerized monthly, quarterly, and 12-month sales projectionsThese projections are aggregated by region, then globally, at Tupperware’s World HeadquartersTupperware uses all techniques discussed in text7© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Three Key Factors for TupperwareThe number of registered “consultants” or sales representativesThe percentage of currently “active” dealers (this number changes each week and month)Sales per active dealer, on a weekly basis8© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Tupperware - Forecast by ConsensusAlthough inputs come from sales, marketing, finance, and production, final forecasts are the consensus of all participating managers.The final step is Tupperware’s version of the “jury of executive opinion”9© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458What is Forecasting?Process of predicting a future eventUnderlying basis of all business decisionsProductionInventoryPersonnelFacilitiesSales will be $200 Million!10© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Short-range forecastUp to 1 year; usually less than 3 monthsJob scheduling, worker assignmentsMedium-range forecast3 months to 3 yearsSales & production planning, budgetingLong-range forecast3+ yearsNew product planning, facility locationTypes of Forecasts by Time Horizon11© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Short-term vs. Longer-term ForecastingMedium/long range forecasts deal with more comprehensive issues and support management decisions regarding planning and products, plants and processes.Short-term forecasting usually employs different methodologies than longer-term forecastingShort-term forecasts tend to be more accurate than longer-term forecasts.12© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Influence of Product Life CycleStages of introduction and growth require longer forecasts than maturity and declineForecasts useful in projecting staffing levels, inventory levels, and factory capacity as product passes through life cycle stages Introduction, Growth, Maturity, Decline13© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Strategy and Issues During a Product’s LifeIntroductionGrowthMaturityDeclineStandardizationLess rapid product changes - more minor changesOptimum capacityIncreasing stability of processLong production runsProduct improvement and cost cuttingLittle product differentiationCost minimizationOver capacity in the industryPrune line to eliminate items not returning good marginReduce capacityForecasting criticalProduct and process reliabilityCompetitive product improvements and optionsIncrease capacityShift toward product focusedEnhance distributionProduct design and development criticalFrequent product and process design changesShort production runsHigh production costsLimited modelsAttention to qualityBest period to increase market shareR&D product engineering criticalPractical to change price or quality imageStrengthen nicheCost control criticalPoor time to change image, price, or qualityCompetitive costs become criticalDefend market positionOM Strategy/IssuesCompany Strategy/IssuesHDTVCD-ROMColor copiersDrive-thru restaurantsFax machinesStation wagonsSales3 1/2” Floppy disksInternet14© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Types of ForecastsEconomic forecastsAddress business cycle, e.g., inflation rate, money supply etc.Technological forecastsPredict rate of technological progressPredict acceptance of new productDemand forecastsPredict sales of existing product15© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Seven Steps in ForecastingDetermine the use of the forecastSelect the items to be forecastedDetermine the time horizon of the forecastSelect the forecasting model(s)Gather the dataMake the forecastValidate and implement results16© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Product Demand Charted over 4 Years with Trend and SeasonalityYear1Year2Year3Year4Seasonal peaksTrend componentActual demand lineAverage demand over four yearsDemand for product or serviceRandom variation17© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Actual Demand, Moving Average, Weighted Moving AverageActual salesMoving averageWeighted moving average18© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Realities of ForecastingForecasts are seldom perfectMost forecasting methods assume that there is some underlying stability in the systemBoth product family and aggregated product forecasts are more accurate than individual product forecasts19© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Forecasting ApproachesUsed when situation is ‘stable’ & historical data existExisting productsCurrent technologyInvolves mathematical techniquese.g., forecasting sales of color televisionsQuantitative MethodsUsed when situation is vague & little data existNew productsNew technologyInvolves intuition, experiencee.g., forecasting sales on InternetQualitative Methods20© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Overview of Qualitative MethodsJury of executive opinionPool opinions of high-level executives, sometimes augment by statistical modelsDelphi methodPanel of experts, queried iterativelySales force compositeEstimates from individual salespersons are reviewed for reasonableness, then aggregated Consumer Market SurveyAsk the customer21© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Involves small group of high-level managersGroup estimates demand by working togetherCombines managerial experience with statistical modelsRelatively quick‘Group-think’disadvantage© 1995 Corel Corp.Jury of Executive Opinion22© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Sales Force CompositeEach salesperson projects his or her salesCombined at district & national levelsSales reps know customers’ wantsTends to be overly optimisticSales© 1995 Corel Corp.23© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Delphi MethodIterative group process3 types of peopleDecision makersStaffRespondentsReduces ‘group-think’Respondents Staff Decision Makers(Sales?)(What will sales be? survey)(Sales will be 45, 50, 55)(Sales will be 50!)24© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Consumer Market SurveyAsk customers about purchasing plansWhat consumers say, and what they actually do are often differentSometimes difficult to answerHow many hours will you use the Internet next week?© 1995 Corel Corp.25© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Overview of Quantitative ApproachesNaïve approachMoving averagesExponential smoothingTrend projectionLinear regressionTime-series ModelsAssociative models26© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Quantitative Forecasting Methods (Non-Naive)QuantitativeForecastingLinearRegressionAssociativeModelsExponentialSmoothingMovingAverageTime SeriesModelsTrendProjection27© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458 Set of evenly spaced numerical data Obtained by observing response variable at regular time periodsForecast based only on past valuesAssumes that factors influencing past and present will continue influence in futureExampleYear: 1998 1999 2000 2001 2002Sales: 78.7 63.5 89.7 93.2 92.1 What is a Time Series?28© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458TrendSeasonalCyclicalRandomTime Series Components29© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Persistent, overall upward or downward patternDue to population, technology etc.Several years duration Mo., Qtr., Yr.Response© 1984-1994 T/Maker Co.Trend Component30© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Regular pattern of up & down fluctuationsDue to weather, customs etc.Occurs within 1 year Mo., Qtr.ResponseSummer© 1984-1994 T/Maker Co.Seasonal Component31© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Common Seasonal PatternsPeriod of Pattern“Season” LengthNumber of “Seasons” in PatternWeekDay7MonthWeek4 – 4 ½MonthDay28 – 31YearQuarter4YearMonth12YearWeek5232© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Repeating up & down movementsDue to interactions of factors influencing economyUsually 2-10 years duration Mo., Qtr., Yr.ResponseCycleCyclical Component33© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Erratic, unsystematic, ‘residual’ fluctuationsDue to random variation or unforeseen eventsUnion strikeTornadoShort duration & nonrepeating © 1984-1994 T/Maker Co.Random Component34© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Any observed value in a time series is the product (or sum) of time series componentsMultiplicative modelYi = Ti · Si · Ci · Ri (if quarterly or mo. data)Additive modelYi = Ti + Si + Ci + Ri (if quarterly or mo. data)General Time Series Models35© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Naive ApproachAssumes demand in next period is the same as demand in most recent periode.g., If May sales were 48, then June sales will be 48Sometimes cost effective & efficient© 1995 Corel Corp.36© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458 MA is a series of arithmetic means Used if little or no trend Used often for smoothingProvides overall impression of data over timeEquationMAnnDemand in Previous PeriodsMoving Average Method37© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458You’re manager of a museum store that sells historical replicas. You want to forecast sales (000) for 2003 using a 3-period moving average. 1998 4 1999 6 2000 5 2001 3 2002 7© 1995 Corel Corp.Moving Average Example38© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Moving Average Solution39© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Moving Average Solution40© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Moving Average Solution41© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458959697989900YearSales2468ActualForecastMoving Average Graph42© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Used when trend is present Older data usually less importantWeights based on intuitionOften lay between 0 & 1, & sum to 1.0EquationWMA =Σ(Weight for period n) (Demand in period n) ΣWeightsWeighted Moving Average Method43© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Actual Demand, Moving Average, Weighted Moving AverageActual salesMoving averageWeighted moving average44© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Increasing n makes forecast less sensitive to changesDo not forecast trend wellRequire much historical data© 1984-1994 T/Maker Co.Disadvantages of Moving Average Methods45© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Form of weighted moving averageWeights decline exponentiallyMost recent data weighted mostRequires smoothing constant ()Ranges from 0 to 1Subjectively chosenInvolves little record keeping of past dataExponential Smoothing Method46© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = At - 1 + (1-)At - 2 + (1- )2·At - 3  + (1- )3At - 4 + ... + (1- )t-1·A0Ft = Forecast value At = Actual value = Smoothing constantFt = Ft-1 + (At-1 - Ft-1)Use for computing forecastExponential Smoothing Equations47© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458During the past 8 quarters, the Port of Baltimore has unloaded large quantities of grain. ( = .10). The first quarter forecast was 175.. Quarter Actual 1 180 2 168 3 159 4 175 5 190 6 205 7 180 8 182 9 ?Exponential Smoothing ExampleFind the forecast for the 9th quarter.48© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1) QuarterActualForecast, Ft(α= .10)1180175.00 (Given)21683159417551906205175.00 +Exponential Smoothing Solution49© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(3159417551906205Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)50© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 -3159417551906205Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)51© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 - 175.00)3159417551906205Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)52© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 - 175.00) = 175.503159417551906205Exponential Smoothing SolutionFt = Ft-1 + 0.1(At-1 - Ft-1)53© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1) QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 - 175.00) = 175.503159175.50 + .10(168 - 175.50) = 174.75417551906205Exponential Smoothing Solution54© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1)QuarterActualForecast, Ft(α= .10)1995180175.00 (Given)1996168175.00 + .10(180 - 175.00) = 175.501997159175.50 + .10(168 - 175.50) = 174.75199817519991902000205174.75 + .10(159 - 174.75)= 173.18Exponential Smoothing Solution55© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1)QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 - 175.00) = 175.503159175.50 + .10(168 - 175.50) = 174.754175174.75 + .10(159 - 174.75) = 173.185190173.18 + .10(175 - 173.18) = 173.366205Exponential Smoothing Solution56© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1)QuarterActualForecast, Ft(α= .10)1180175.00 (Given)2168175.00 + .10(180 - 175.00) = 175.503159175.50 + .10(168 - 175.50) = 174.754175174.75 + .10(159 - 174.75) = 173.185190173.18 + .10(175 - 173.18) = 173.366205173.36 + .10(190 - 173.36) = 175.02Exponential Smoothing Solution57© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1)TimeActualForecast, Ft(α= .10)4175174.75 + .10(159 - 174.75) = 173.185190173.18 + .10(175 - 173.18) = 173.366205173.36 + .10(190 - 173.36) = 175.02Exponential Smoothing Solution71808175.02 + .10(205 - 175.02) = 178.02958© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Ft-1 + 0.1(At-1 - Ft-1)TimeActualForecast, Ft(α= .10)4175174.75 + .10(159 - 174.75) = 173.185190173.18 + .10(175 - 173.18) = 173.366205173.36 + .10(190 - 173.36) = 175.02Exponential Smoothing Solution71808175.02 + .10(205 - 175.02) = 178.029178.22 + .10(182 - 178.22) = 178.58 182178.02 + .10(180 - 178.02) = 178.22?59© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- )At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%60© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- ) At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%9%61© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- )At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%9%8.1%62© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- )At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%9%8.1%90%63© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- ) At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%9%8.1%90%9%64© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft =  At - 1 + (1- ) At - 2 + (1- )2At - 3 + ...Forecast Effects of Smoothing Constant WeightsPrior Period2 periods ago(1 - )3 periods ago(1 - )2== 0.10= 0.9010%9%8.1%90%9%0.9%65© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Impact of 66© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Choosing Seek to minimize the Mean Absolute Deviation (MAD)If: Forecast error = demand - forecastThen:67© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Exponential Smoothing with Trend AdjustmentForecast including trend (FITt) = exponentially smoothed forecast (Ft) + exponentially smoothed trend (Tt)68© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = Last period’s forecast + (Last period’s actual – Last period’s forecast)Ft = Ft-1 +  (At-1 – Ft-1)orTt = (Forecast this period - Forecast last period) + (1-)(Trend estimate last periodTt = (Ft - Ft-1) + (1- )Tt-1 orExponential Smoothing with Trend Adjustment - continued69© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Ft = exponentially smoothed forecast of the data series in period tTt = exponentially smoothed trend in period tAt = actual demand in period t = smoothing constant for the average = smoothing constant for the trendExponential Smoothing with Trend Adjustment - continued70© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Comparing Actual and Forecasts71© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Regression72© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Least SquaresDeviationDeviationDeviationDeviationDeviationDeviationDeviationTimeValues of Dependent VariableActual observationPoint on regression line73© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Actual and the Least Squares Line74© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Used for forecasting linear trend lineAssumes relationship between response variable, Y, and time, X, is a linear functionEstimated by least squares methodMinimizes sum of squared errorsiYabXi=+Linear Trend Projection75© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458b > 0b < 0aaYTime, XLinear Trend Projection Model76© 2004 by Prentice Hall, Inc., Upper Saddle River, N.J. 07458Scatter Di