Slide 1 of 56 Chapter 3 Chapter 3 Forecasting in POM: Forecasting in POM: The Starting Point for All The Starting Point for All Planning Planning
Mar 29, 2015
Slide 1 of 56
Chapter 3Chapter 3Chapter 3Chapter 3
Forecasting in POM:Forecasting in POM:
The Starting Point for All PlanningThe Starting Point for All Planning
Slide 2 of 56
OverviewOverviewOverviewOverview
IntroductionIntroduction Qualitative Forecasting MethodsQualitative Forecasting Methods Quantitative Forecasting ModelsQuantitative Forecasting Models How to Have a Successful Forecasting SystemHow to Have a Successful Forecasting System Computer Software for ForecastingComputer Software for Forecasting Forecasting in Small Businesses and Start-Up Forecasting in Small Businesses and Start-Up
VenturesVentures Wrap-Up: What World-Class Producers DoWrap-Up: What World-Class Producers Do
Slide 3 of 56
Demand ManagementDemand ManagementDemand ManagementDemand Management
Independent demand items are the only Independent demand items are the only items demand for which needs to be items demand for which needs to be forecastforecast
These items include:These items include: Finished goods andFinished goods and Spare partsSpare parts
Slide 4 of 56
Demand ManagementDemand ManagementDemand ManagementDemand Management
A
Independent Demand(finished goods and spare parts)
B(4) C(2)
D(2) E(1) D(3) F(2)
Dependent Demand(components)
Slide 5 of 56
Introduction Introduction Introduction Introduction
Demand estimatesDemand estimates for independent demand products for independent demand products and services are the starting point for all the other and services are the starting point for all the other forecasts in POM.forecasts in POM.
Management teams develop Management teams develop sales forecastssales forecasts based in based in part on demand estimates.part on demand estimates.
Sales forecasts become inputs to both business Sales forecasts become inputs to both business strategy and strategy and production resource forecastsproduction resource forecasts..
Slide 6 of 56
Forecasting is an Integral PartForecasting is an Integral Part of Business Planning of Business Planning
Forecasting is an Integral PartForecasting is an Integral Part of Business Planning of Business Planning
ForecastForecastMethod(s)Method(s)
DemandDemandEstimatesEstimates
SalesSalesForecastForecast
ManagementManagementTeamTeam
Inputs:Inputs:Market,Market,
Economic,Economic,OtherOther
BusinessBusinessStrategyStrategy
Production ResourceProduction ResourceForecastsForecasts
Slide 7 of 56
Examples of Production Resource ForecastsExamples of Production Resource ForecastsExamples of Production Resource ForecastsExamples of Production Resource Forecasts
Forecast Forecast HorizonHorizon
Time SpanTime Span Item Being ForecastItem Being ForecastUnits of Units of MeasureMeasure
Long-RangeLong-Range YearsYears
Product linesProduct lines Factory capacitiesFactory capacities Planning for new productsPlanning for new products Capital expendituresCapital expenditures Facility location or expansion Facility location or expansion R&DR&D
Dollars, tons, etc.Dollars, tons, etc.
Medium-Medium-RangeRange
MonthsMonths
Product groupsProduct groups Department capacitiesDepartment capacities Sales planningSales planning Production planning and Production planning and budgetingbudgeting
Dollars, tons, etc.Dollars, tons, etc.
Short-RangeShort-Range WeeksWeeks
Specific product quantitiesSpecific product quantities Machine capacities Machine capacities PlanningPlanning PurchasingPurchasing SchedulingScheduling Workforce levelsWorkforce levels Production levelsProduction levels Job assignmentsJob assignments
Physical units of Physical units of productsproducts
Slide 8 of 56
Forecasting MethodsForecasting MethodsForecasting MethodsForecasting Methods
Qualitative ApproachesQualitative Approaches Quantitative ApproachesQuantitative Approaches
Slide 9 of 56
Qualitative Forecasting ApplicationsQualitative Forecasting ApplicationsSmall and Large FirmsSmall and Large Firms
Qualitative Forecasting ApplicationsQualitative Forecasting ApplicationsSmall and Large FirmsSmall and Large Firms
TechniqueTechnique Low Sales Low Sales (less than $100M)(less than $100M)
High SalesHigh Sales(more than $500M)(more than $500M)
Manager’s OpinionManager’s Opinion 40.7%40.7% 39.6%39.6%
Executive’s Executive’s OpinionOpinion
40.7%40.7% 41.6%41.6%
Sales Force Sales Force CompositeComposite
29.6%29.6% 35.4%35.4%
Number of FirmsNumber of Firms 2727 4848
Source: Nada Sanders and Karl Mandrodt (1994) “Practitioners Continue to Rely on Judgmental Forecasting Methods Instead of Quantitative Methods,” Interfaces, vol. 24, no. 2, pp. 92-100.Note: More than one response was permitted.
Slide 10 of 56
Qualitative ApproachesQualitative ApproachesQualitative ApproachesQualitative Approaches
Usually based on judgments about causal factors that Usually based on judgments about causal factors that underlie the demand of particular products or servicesunderlie the demand of particular products or services
Do not require a demand history for the product or Do not require a demand history for the product or service, therefore are useful for new products/servicesservice, therefore are useful for new products/services
Approaches vary in sophistication from scientifically Approaches vary in sophistication from scientifically conducted surveys to intuitive hunches about future conducted surveys to intuitive hunches about future eventsevents
Slide 11 of 56
Qualitative Methods Qualitative Methods Qualitative Methods Qualitative Methods
Executive committee consensusExecutive committee consensus Delphi methodDelphi method Survey of sales forceSurvey of sales force Survey of customersSurvey of customers Historical analogyHistorical analogy Market researchMarket research
Slide 12 of 56
Quantitative Forecasting ApproachesQuantitative Forecasting ApproachesQuantitative Forecasting ApproachesQuantitative Forecasting Approaches
Based on the assumption that the “forces” that Based on the assumption that the “forces” that generated the past demand will generate the future generated the past demand will generate the future demand, i.e., history will tend to repeat itselfdemand, i.e., history will tend to repeat itself
Analysis of the past demand pattern provides a good Analysis of the past demand pattern provides a good basis for forecasting future demandbasis for forecasting future demand
Majority of quantitative approaches fall in the Majority of quantitative approaches fall in the category of time series analysiscategory of time series analysis
Slide 13 of 56
Quantitative Forecasting ApplicationsQuantitative Forecasting ApplicationsSmall and Large FirmsSmall and Large Firms
Quantitative Forecasting ApplicationsQuantitative Forecasting ApplicationsSmall and Large FirmsSmall and Large Firms
TechniqueTechnique Low Sales Low Sales (less than $100M)(less than $100M)
High SalesHigh Sales(more than $500M)(more than $500M)
Moving AverageMoving Average 29.6%29.6% 29.229.2
Simple Linear Simple Linear RegressionRegression
14.8%14.8% 14.614.6
NaiveNaive 18.5%18.5% 14.614.6
Single Exponential Single Exponential SmoothingSmoothing
14.8%14.8% 20.820.8
Multiple RegressionMultiple Regression 22.2%22.2% 27.127.1
SimulationSimulation 3.7%3.7% 10.410.4
Classical DecompositionClassical Decomposition 3.7%3.7% 8.38.3
Box-JenkinsBox-Jenkins 3.7%3.7% 6.36.3
Number of FirmsNumber of Firms 2727 4848
Source: Nada Sanders and Karl Mandrodt (1994) “Practitioners Continue to Rely on Judgmental Forecasting Methods Instead of Quantitative Methods,” Interfaces, vol. 24, no. 2, pp. 92-100.Note: More than one response was permitted.
Slide 14 of 56
A A time seriestime series is a set of numbers where the order or is a set of numbers where the order or sequence of the numbers is important, e.g., historical sequence of the numbers is important, e.g., historical demanddemand
Analysis of the time series identifies patternsAnalysis of the time series identifies patterns Once the patterns are identified, they can be used to Once the patterns are identified, they can be used to
develop a forecastdevelop a forecast
Time Series AnalysisTime Series AnalysisTime Series AnalysisTime Series Analysis
Slide 15 of 56
Components of Time SeriesComponents of Time SeriesComponents of Time SeriesComponents of Time Series
1 2 3 4
x
x xx
xx
x xx
xx x x x
xxxxxx x x
xx
x x xx
xx
xx
x
xx
xx
xx
xx
xx
xx
x
x
Year
Sal
es
What’s going on here?
Slide 16 of 56
Components of Time Series Components of Time Series Components of Time Series Components of Time Series
TrendsTrends are noted by an upward or downward sloping are noted by an upward or downward sloping lineline
SeasonalitySeasonality is a data pattern that repeats itself over is a data pattern that repeats itself over the period of one year or less the period of one year or less
CycleCycle is a data pattern that repeats itself... may take is a data pattern that repeats itself... may take yearsyears
Irregular variationsIrregular variations are jumps in the level of the series are jumps in the level of the series due to extraordinary eventsdue to extraordinary events
Random fluctuationRandom fluctuation from random variation or from random variation or unexplained causesunexplained causes
Slide 17 of 56
Actual Data & the Regression LineActual Data & the Regression LineActual Data & the Regression LineActual Data & the Regression Line
40
60
80
100
120
140
160
1988 1989 1990 1991 1992 1993 1994 1995 1996 1997
Year
Pow
er D
eman
d
Actual Data
Linear (Actual Data)
l
Slide 18 of 56
SeasonalitySeasonalitySeasonalitySeasonality
Length of TimeLength of Time Number ofNumber of
Before PatternBefore Pattern Length ofLength of SeasonsSeasons
Is RepeatedIs Repeated SeasonSeason in Patternin Pattern
YearYear QuarterQuarter 4 4
YearYear Month Month 1212
YearYear Week Week 5252
MonthMonth Week Week 4 4
MonthMonth Day Day 28-31 28-31
WeekWeek Day Day 7 7
Slide 19 of 56
Eight Steps to ForecastingEight Steps to ForecastingEight Steps to ForecastingEight Steps to Forecasting
Determining the use of the forecast--what are the Determining the use of the forecast--what are the objectives?objectives?
Select the items to be forecastSelect the items to be forecast Determine the time horizon of the forecastDetermine the time horizon of the forecast Select the forecasting model(s)Select the forecasting model(s) Collect the dataCollect the data Validate the forecasting modelValidate the forecasting model Make the forecastMake the forecast Implement the resultsImplement the results
Slide 20 of 56
Quantitative Forecasting ApproachesQuantitative Forecasting ApproachesQuantitative Forecasting ApproachesQuantitative Forecasting Approaches
Linear RegressionLinear Regression Simple Moving AverageSimple Moving Average Weighted Moving AverageWeighted Moving Average Exponential Smoothing (exponentially weighted Exponential Smoothing (exponentially weighted
moving average)moving average) Exponential Smoothing with Trend (double Exponential Smoothing with Trend (double
smoothing)smoothing)
Slide 21 of 56
Simple Linear RegressionSimple Linear RegressionSimple Linear RegressionSimple Linear Regression
Relationship between one independent variable, X, Relationship between one independent variable, X, and a dependent variable, Y.and a dependent variable, Y.
Assumed to be linear (a straight line)Assumed to be linear (a straight line) Form: Y = a + bXForm: Y = a + bX
Y = dependent variableY = dependent variable X = independent variableX = independent variable a = y-axis intercepta = y-axis intercept b = slope of regression lineb = slope of regression line
Slide 22 of 56
Simple Linear Regression ModelSimple Linear Regression ModelSimple Linear Regression ModelSimple Linear Regression Model
b is similar to the slope. However, since it is b is similar to the slope. However, since it is calculated with the variability of the data in mind, calculated with the variability of the data in mind, its formulation is not as straight-forward as our its formulation is not as straight-forward as our usual notion of slope usual notion of slope
Yt = a + bx
0 1 2 3 4 5 x (weeks)
Y
Slide 23 of 56
Calculating a and bCalculating a and bCalculating a and bCalculating a and b
a = y - bx
b =xy - n(y)(x)
x - n(x2 2
)
Slide 24 of 56
Regression Equation ExampleRegression Equation ExampleRegression Equation ExampleRegression Equation Example
Week Sales1 1502 1573 1624 1665 177
Develop a regression equation to predict sales based on these five points.
Week Week*Week Sales Week*Sales1 1 150 1502 4 157 3143 9 162 4864 16 166 6645 25 177 8853 55 162.4 2499
Average Sum Average Sum
b =xy - n(y)(x)
x - n(x=
2499 - 5(162.4)(3)=
a = y - bx = 162.4 - (6.3)(3) =
2 2
) ( )55 5 9
63
106.3
143.5
Regression Equation ExampleRegression Equation Example
Slide 25 of 55
y = 143.5 + 6.3t
135140145150155
160165170175180
1 2 3 4 5 Period
Sal
es
Sales
Forecast
Regression Equation ExampleRegression Equation Example
Slide 26 of 55
Slide 27 of 56
Forecast AccuracyForecast AccuracyForecast AccuracyForecast Accuracy
Accuracy is the typical criterion for judging the Accuracy is the typical criterion for judging the performance of a forecasting approachperformance of a forecasting approach
Accuracy is how well the forecasted values match the Accuracy is how well the forecasted values match the actual valuesactual values
Slide 28 of 56
Monitoring Accuracy Monitoring Accuracy Monitoring Accuracy Monitoring Accuracy
Accuracy of a forecasting approach needs to be Accuracy of a forecasting approach needs to be monitored to assess the confidence you can have in monitored to assess the confidence you can have in its forecasts and changes in the market may require its forecasts and changes in the market may require reevaluation of the approachreevaluation of the approach
Accuracy can be measured in several waysAccuracy can be measured in several ways Mean absolute deviation (MAD)Mean absolute deviation (MAD) Mean squared error (MSE)Mean squared error (MSE)
Slide 29 of 56
Mean Absolute Deviation (MAD)Mean Absolute Deviation (MAD)Mean Absolute Deviation (MAD)Mean Absolute Deviation (MAD)
n
demandForecast -demand Actual=MAD
n
1=i i
n
)F - (An
1
ii iMAD
n
)F - (An
1
ii iMAD
Slide 30 of 56
Mean Squared Error (MSE)Mean Squared Error (MSE)Mean Squared Error (MSE)Mean Squared Error (MSE)
MSE = (SMSE = (Syxyx))22
Small value for SSmall value for Syxyx means data points tightly means data points tightly
grouped around the line and error range is small. grouped around the line and error range is small. The smaller the standard error the more accurate The smaller the standard error the more accurate the forecast.the forecast.
MSE = 1.25(MAD)MSE = 1.25(MAD)
When the forecast errors are normally distributedWhen the forecast errors are normally distributed
Slide 31 of 56
Example--MADExample--MADExample--MADExample--MAD
Month Sales Forecast1 220 n/a2 250 2553 210 2054 300 3205 325 315
Determine the MAD for the four forecast periodsDetermine the MAD for the four forecast periods
Slide 32 of 56
SolutionSolutionSolutionSolution
MAD = A - F
n=
40
4= 10
t tt=1
n
Month Sales Forecast Abs Error1 220 n/a2 250 255 53 210 205 54 300 320 205 325 315 10
40
Slide 33 of 56
Simple Moving AverageSimple Moving AverageSimple Moving AverageSimple Moving Average
An averaging period (AP) is given or selectedAn averaging period (AP) is given or selected The forecast for the next period is the arithmetic The forecast for the next period is the arithmetic
average of the AP most recent actual demandsaverage of the AP most recent actual demands It is called a “simple” average because each period It is called a “simple” average because each period
used to compute the average is equally weightedused to compute the average is equally weighted . . . more. . . more
Slide 34 of 56
Simple Moving AverageSimple Moving AverageSimple Moving AverageSimple Moving Average
It is called “moving” because as new demand data It is called “moving” because as new demand data becomes available, the oldest data is not usedbecomes available, the oldest data is not used
By increasing the AP, the forecast is less responsive By increasing the AP, the forecast is less responsive to fluctuations in demand (low impulse response)to fluctuations in demand (low impulse response)
By decreasing the AP, the forecast is more responsive By decreasing the AP, the forecast is more responsive to fluctuations in demand (high impulse response)to fluctuations in demand (high impulse response)
Slide 35 of 56
Simple Moving AverageSimple Moving AverageSimple Moving AverageSimple Moving Average
Week Demand1 6502 6783 7204 7855 8596 9207 8508 7589 892
10 92011 78912 844
F = A + A + A +...+A
ntt-1 t-2 t-3 t-n
Let’s develop 3-week and 6-Let’s develop 3-week and 6-week moving average forecasts week moving average forecasts for demand. for demand.
Assume you only have 3 weeks Assume you only have 3 weeks and 6 weeks of actual demand and 6 weeks of actual demand data for the respective forecastsdata for the respective forecasts
Week Demand 3-Week 6-Week1 6502 6783 7204 785 682.675 859 727.676 920 788.007 850 854.67 768.678 758 876.33 802.009 892 842.67 815.33
10 920 833.33 844.0011 789 856.67 866.5012 844 867.00 854.83
Simple Moving AverageSimple Moving AverageSimple Moving AverageSimple Moving Average
Slide 36 of 55
500
600
700
800
900
1000
1 2 3 4 5 6 7 8 9 10 11 12
Week
Dem
and Demand
3-Week
6-Week
Simple Moving AverageSimple Moving AverageSimple Moving AverageSimple Moving Average
Slide 37 of 55
Slide 38 of 56
Weighted Moving AverageWeighted Moving AverageWeighted Moving AverageWeighted Moving Average
This is a variation on the simple moving average This is a variation on the simple moving average where instead of the weights used to compute the where instead of the weights used to compute the average being equal, they are not equalaverage being equal, they are not equal
This allows more recent demand data to have a This allows more recent demand data to have a greater effect on the moving average, therefore the greater effect on the moving average, therefore the forecastforecast
. . . more. . . more
Slide 39 of 56
Weighted Moving AverageWeighted Moving AverageWeighted Moving AverageWeighted Moving Average
The weights must add to 1.0 and generally decrease The weights must add to 1.0 and generally decrease in value with the age of the datain value with the age of the data
The distribution of the weights determine impulse The distribution of the weights determine impulse response of the forecastresponse of the forecast
Slide 40 of 56
Weighted Moving AverageWeighted Moving AverageWeighted Moving AverageWeighted Moving Average
F = w A + w A + w A +...+w At 1 t-1 2 t-2 3 t-3 n t-n
w = 1ii=1
n
Determine the 3-period weighted moving average forecast for period 4
Weights (adding up to 1.0): t-1: .5t-2: .3t-3: .2
Week Demand1 6502 6783 7204
Slide 41 of 56
SolutionSolutionSolutionSolution
Week Demand Forecast1 6502 6783 7204 693.4
F= .5(720)+.3(678)+.2(650)4
Slide 42 of 56
Exponential SmoothingExponential SmoothingExponential SmoothingExponential Smoothing
The weights used to compute the forecast (moving The weights used to compute the forecast (moving average) are exponentially distributedaverage) are exponentially distributed
The forecast is the sum of the old forecast and a The forecast is the sum of the old forecast and a portion of the forecast errorportion of the forecast error
FFtt = F = Ft-1t-1 + + (A(At-1t-1--FFt-1t-1)) . . . more. . . more
Slide 43 of 56
Exponential SmoothingExponential SmoothingExponential SmoothingExponential Smoothing
The smoothing constant, The smoothing constant, , must be between 0.0 and , must be between 0.0 and 1.0 (excluding 0.0 and 1.0)1.0 (excluding 0.0 and 1.0)
A large A large provides a high impulse response forecast provides a high impulse response forecast A small A small provides a low impulse response forecast provides a low impulse response forecast
Slide 44 of 56
Exponential Smoothing ExampleExponential Smoothing ExampleExponential Smoothing ExampleExponential Smoothing Example
Week Demand1 8202 7753 6804 6555 7506 8027 7988 6899 775
10
Determine exponential Determine exponential smoothing forecasts for smoothing forecasts for periods 2 through 10 periods 2 through 10 using using =.10 and =.10 and =.60.=.60.
Let FLet F11=D=D1 1
Week Demand 0.1 0.61 820 820.00 820.002 775 820.00 820.003 680 815.50 820.004 655 801.95 817.305 750 787.26 808.096 802 783.53 795.597 798 785.38 788.358 689 786.64 786.579 775 776.88 786.61
10 776.69 780.77
Exponential Smoothing ExampleExponential Smoothing ExampleExponential Smoothing ExampleExponential Smoothing Example
Slide 45 of 55
Slide 46 of 56
Effect of Effect of on Forecast on ForecastEffect of Effect of on Forecast on Forecast
500
600
700
800
900
1 2 3 4 5 6 7 8 9 10
Week
Dem
and
Demand
0.1
0.6
Slide 47 of 56
Criteria for SelectingCriteria for Selectinga Forecasting Methoda Forecasting MethodCriteria for SelectingCriteria for Selectinga Forecasting Methoda Forecasting Method
CostCost AccuracyAccuracy Data availableData available Time spanTime span Nature of products and servicesNature of products and services Impulse response and noise dampeningImpulse response and noise dampening
Slide 48 of 56
Reasons for Ineffective ForecastingReasons for Ineffective ForecastingReasons for Ineffective ForecastingReasons for Ineffective Forecasting
Not involving a broad cross section of peopleNot involving a broad cross section of people Not recognizing that forecasting is integral to Not recognizing that forecasting is integral to
business planningbusiness planning Not recognizing that forecasts will always be wrong Not recognizing that forecasts will always be wrong
(think in terms of interval rather than point forecasts)(think in terms of interval rather than point forecasts) Not forecasting the right things Not forecasting the right things
(forecast independent demand only)(forecast independent demand only) Not selecting an appropriate forecasting method Not selecting an appropriate forecasting method
(use MAD to evaluate goodness of fit)(use MAD to evaluate goodness of fit) Not tracking the accuracy of the forecasting modelsNot tracking the accuracy of the forecasting models
Slide 49 of 56
How to Monitor andHow to Monitor andControl a Forecasting ModelControl a Forecasting Model
How to Monitor andHow to Monitor andControl a Forecasting ModelControl a Forecasting Model
Tracking SignalTracking Signal
Tracking signal = Tracking signal =
= =
MAD
demand)Forecast - demand (Actualn
1
ii
MAD
demand)Forecast - demand (Actualn
1
ii
MAD
)F - (An
1
iii
MAD
)F - (An
1
iii
Slide 50 of 56
Tracking Signal: What do you notice?Tracking Signal: What do you notice?Tracking Signal: What do you notice?Tracking Signal: What do you notice?
20
25
30
35
40
0 1 2 3 4 5 6 7 8 9 10 11
Period
Sal
es
Slide 51 of 56
Sources of Forecasting DataSources of Forecasting DataSources of Forecasting DataSources of Forecasting Data
Consumer Confidence IndexConsumer Confidence Index Consumer Price IndexConsumer Price Index Housing StartsHousing Starts Index of Leading Economic IndicatorsIndex of Leading Economic Indicators Personal Income and ConsumptionPersonal Income and Consumption Producer Price IndexProducer Price Index Purchasing Manager’s IndexPurchasing Manager’s Index Retail SalesRetail Sales
Slide 52 of 56
Wrap-Up: World-Class PracticeWrap-Up: World-Class PracticeWrap-Up: World-Class PracticeWrap-Up: World-Class Practice
Predisposed to have effective methods of forecasting Predisposed to have effective methods of forecasting because they have exceptional long-range business because they have exceptional long-range business planningplanning
Formal forecasting effortFormal forecasting effort Develop methods to monitor the performance of their Develop methods to monitor the performance of their
forecasting modelsforecasting models Use forecasting software with automated model Use forecasting software with automated model
fitting features, which is readily available todayfitting features, which is readily available today Do not overlook the short run.... excellent short range Do not overlook the short run.... excellent short range
forecasts as wellforecasts as well