Chapter 4 Forecasting Production Planning Overview What is forecasting? Types of forecasts 7 steps of forecasting Qualitative forecasting.

Chapter 4 Forecasting

Production Planning

SalesForecast

AggregatePlanning

MasterProductionSchedule

ProductionScheduling

Purchasing

Production

Overview

What is forecasting?

Types of forecasts

7 steps of forecasting

Qualitative forecasting

Overview

Quantitative forecastingTime-series forecasting

NaïveMoving averageExponential smoothingSeasonal variations

Associative methods Monitoring and

Controlling Forecasts

What is forecasting?

Sales will be $200 Million!Could be a

prediction based on historical data and mathematical models

Could be a prediction based on expertise and intuition

Forecasting - Is the art and science of predicting the future

Could be a prediction based on both a model and a manager’s expertise

7 Steps to a Forecast

Determine the use of the forecast

Select the items to be forecast

Determine the time horizon of the forecast

Select the forecasting model(s)

Gather the data

Make the forecast

Validate and implement results

Realities of Forecasting

Forecasts never perfect and seldom correct.

Most forecasting methods assume that there is

some underlying stability in the system

Both product family and aggregated product

forecasts are more accurate than individual product

forecasts

Demand Forecasts OM manager is primarily interested in demand

forecasts (as opposed to economic forecasts and technological forecasts)

Underlying basis of all business decisionsProduction InventoryPersonnelFacilities

Demand Forecast ApplicationsDemand Forecast ApplicationsTime Horizon

Medium Term Long Term Short Term (3 months– (more than

Application (0–3 months) 3 years) 3 years)

Forecast quantity

Decision area

Forecastingtechnique

Forecast quantity Individualproducts orservices

Decision area Inventorymanagement

Final assemblyscheduling

Workforcescheduling

Master productionscheduling

Forecasting Time seriestechnique Associative

Forecast quantity Individual Total salesproducts or Groups or familiesservices of products or

servicesDecision area Inventory Staff planning

management ProductionFinal assembly planning

scheduling Master productionWorkforce scheduling

scheduling PurchasingMaster production Distribution

schedulingForecasting Time series Associative

technique Associative

Forecast quantity Individual Total sales Total salesproducts or Groups or familiesservices of products or

servicesDecision area Inventory Staff planning Facility location

management Production CapacityFinal assembly planning planning

scheduling Master production ProcessWorkforce scheduling management

scheduling PurchasingMaster production Distribution

schedulingForecasting Time series Associative Associative

technique Associative

Overview of Qualitative MethodsOverview of Qualitative MethodsREAD in TEXT (p. 81-82)

Jury of executive opinion Pool opinions of high-level executives, sometimes augment by statistical

models

Sales force composite Estimates from individual salespersons are reviewed for reasonableness, then

aggregated

Delphi method Panel of experts, queried iteratively

Consumer Market Survey Ask the customer

Patterns of DemandPatterns of Demand

Patterns of DemandPatterns of DemandQ

(a) Random: Data cluster about a horizontal line.

(b) Trend: Data consistently increase or decrease over a period of time.

Months

(c) Seasonal: Data consistently show peaks and valleys at the same time each year.

Year 1

Months

Year 1

Year 2

(c) Seasonal: Data consistently show peaks and valleys at the same time each year.

| | | | | |1 2 3 4 5 6

(c) Cyclical: Data reveal gradual increases and decreases over extended periods.

Overview of Quantitative Methods

Naïve approach

Moving averages

Exponential smoothing

Linear regression

Time-series Models – no trend, seasonal, or cyclical fluctuations

Associative models

Set of evenly spaced numerical data Obtained by observing response variable at regular time

periods

Forecast based only on past values Assumes that factors influencing past and present will

continue influence in future

ExampleYear: 1993 1994 1995 1996 1997

Sales: 78.7 63.5 89.7 93.2 92.1

What is a Time Series?

Naïve Approach

Assumes demand in next period is the same

as demand in most recent periode.g., If May sales were 48, then June sales will

Sometimes cost effective & efficient

Moving Average Approach

MA is a series of arithmetic means

Used if little or no trend

Used often for smoothingProvides overall impression of data over time

MAMAnn

nn Demand inDemand in PreviousPrevious PeriodsPeriods

Time-Series MethodsTime-Series MethodsSimple Moving AveragesSimple Moving Averages

Patient arrivals have been recorded at a medical clinic over the past 28 weeks.

Want to predict the number of patient arrivals for the 29th week.

450 —

430 —

410 —

390 —

370 —Pat

| | | | | |0 5 10 15 20 25 30

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

Actual patientarrivals

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

ls-No trend

-No seasonal variation

-No cycle

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

PatientWeek Arrivals

1 4002 3803 411

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

1 4002 3803 411

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

1 4002 3803 411

F4 = 411 + 380 + 4003

F4 = 397.0

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

1 4002 3803 411

F4 = 397.0

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

2 3803 4114 415

F5 = 415 + 411 + 380

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

2 3803 4114 415

F5 = 402.0

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

Go To Excel

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

3-week MAforecast

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

3-week MAforecast

6-week MAforecast

WEIGHTED Moving Averages

Increasing n makes forecast less sensitive to changes

Do not forecast trend well Require much historical data

Disadvantages of Moving Average Methods

Form of weighted moving averageWeights decline exponentiallyMost recent data weighted most

Requires smoothing constant ()Ranges from 0 to 1Subjectively chosen

Involves little record keeping of past data

Exponential Smoothing Method

Time-Series MethodsTime-Series MethodsExponential SmoothingExponential Smoothing

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

Exponential Smoothing = 0.10

Ft +1 = Ft + (Dt – Ft )

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

F3 = 390 (Given)D3 = 411

Ft +1 = Ft + (Dt – Ft )

F4 = 390 + 0.10(411-390)

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

F4 = 392.1

F3 = 390 (Given)D3 = 411

Ft +1 = Ft + (Dt – Ft )

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

F4 = 392.1D4 = 415

F4 = 392.1 F5 = 394.4

Ft +1 = Ft + (Dt – Ft )

450 —

430 —

410 —

390 —

370 —

| | | | | |0 5 10 15 20 25 30

450 —

430 —

410 —

390 —

370 —Pat

| | | | | |0 5 10 15 20 25 30

Exponential smoothing = 0.10

Go To Excel

450 —

430 —

410 —

390 —

370 —Pat

| | | | | |0 5 10 15 20 25 30

3-week MAforecast

Exponential smoothing = 0.10

Exponential smoothing with trend adjustmentSKIP

Trend projection (p. 93-96)Regression analysis

Time-Series MethodsTime-Series MethodsSeasonal InfluencesSeasonal Influences

Quarter Year 1 Year 2 Year 3 Year 4

1 45 70 100 1002 335 370 585 7253 520 590 830 11604 100 170 285 215

Total 1000 1200 1800 2200 Average 250 300 450 550

Stanley Steamer Carpet Cleaners

0200400600800

100012001400

0 1 2 3 4 5

Quarter

Year 1

Year 2

Year 3

Year 4

1 45 70 100 1002 335 370 585 7253 520 590 830 11604 100 170 285 215

Total 1000 1200 1800 2200 Average 250 300 450 550

Seasonal Index = Actual Demand

Average Demand

Projected Annual Demand = 2600Average Quarterly Demand = 2600/4 = 650

1 45 70 100 1002 335 370 585 7253 520 590 830 11604 100 170 285 215

Total 1000 1200 1800 2200 Average 250 300 450 550

Seasonal Index = Actual Demand

Average Demand

1 45 70 100 1002 335 370 585 7253 520 590 830 11604 100 170 285 215

Total 1000 1200 1800 2200 Average 250 300 450 550

Seasonal Index = = 0.1845

1 45/250 = 0.18 70/300 = 0.23 100/450 = 0.22 100/550 = 0.182 335/250 = 1.34 370/300 = 1.23 585/450 = 1.30 725/550 = 1.323 520/250 = 2.08 590/300 = 1.97 830/450 = 1.84 1160/550 = 2.114 100/250 = 0.40 170/300 = 0.57 285/450 = 0.63 215/550 = 0.39

Quarter Average Seasonal Index

1 (0.18 + 0.23 + 0.22 + 0.18)/4 = 0.20234

1 45/250 = 0.18 70/300 = 0.23 100/450 = 0.22 100/550 = 0.182 335/250 = 1.34 370/300 = 1.23 585/450 = 1.30 725/550 = 1.323 520/250 = 2.08 590/300 = 1.97 830/450 = 1.84 1160/550 = 2.114 100/250 = 0.40 170/300 = 0.57 285/450 = 0.63 215/550 = 0.39

Quarter Average Seasonal Index

1 (0.18 + 0.23 + 0.22 + 0.18)/4 = 0.202 (1.34 + 1.23 + 1.30 + 1.32)/4 = 1.303 (2.08 + 1.97 + 1.84 + 2.11)/4 = 2.004 (0.40 + 0.57 + 0.63 + 0.39)/4 = 0.50

1 45/250 = 0.18 70/300 = 0.23 100/450 = 0.22 100/550 = 0.182 335/250 = 1.34 370/300 = 1.23 585/450 = 1.30 725/550 = 1.323 520/250 = 2.08 590/300 = 1.97 830/450 = 1.84 1160/550 = 2.114 100/250 = 0.40 170/300 = 0.57 285/450 = 0.63 215/550 = 0.39

Quarter Average Seasonal Index Forecast

1 (0.18 + 0.23 + 0.22 + 0.18)/4 = 0.202 (1.34 + 1.23 + 1.30 + 1.32)/4 = 1.303 (2.08 + 1.97 + 1.84 + 2.11)/4 = 2.004 (0.40 + 0.57 + 0.63 + 0.39)/4 = 0.50

1 45/250 = 0.18 70/300 = 0.23 100/450 = 0.22 100/550 = 0.182 335/250 = 1.34 370/300 = 1.23 585/450 = 1.30 725/550 = 1.323 520/250 = 2.08 590/300 = 1.97 830/450 = 1.84 1160/550 = 2.114 100/250 = 0.40 170/300 = 0.57 285/450 = 0.63 215/550 = 0.39

1 (0.18 + 0.23 + 0.22 + 0.18)/4 = 0.20 650(0.20) = 1302 (1.34 + 1.23 + 1.30 + 1.32)/4 = 1.303 (2.08 + 1.97 + 1.84 + 2.11)/4 = 2.004 (0.40 + 0.57 + 0.63 + 0.39)/4 = 0.50

1 45/250 = 0.18 70/300 = 0.23 100/450 = 0.22 100/550 = 0.182 335/250 = 1.34 370/300 = 1.23 585/450 = 1.30 725/550 = 1.323 520/250 = 2.08 590/300 = 1.97 830/450 = 1.84 1160/550 = 2.114 100/250 = 0.40 170/300 = 0.57 285/450 = 0.63 215/550 = 0.39

1 (0.18 + 0.23 + 0.22 + 0.18)/4 = 0.20 650(0.20) = 1302 (1.34 + 1.23 + 1.30 + 1.32)/4 = 1.30 650(1.30) = 8453 (2.08 + 1.97 + 1.84 + 2.11)/4 = 2.00 650(2.00) = 13004 (0.40 + 0.57 + 0.63 + 0.39)/4 = 0.50 650(0.50) = 325

Remember Regression Remember Regression Analysis?Analysis?

Independent variableX

Y Regressionequation:Y = a + bX

Actualvalueof Y

Value of X usedto estimate Y

Regressionequation:Y = a + bX

Actualvalueof Y

Estimate ofY from regressionequation

Actualvalueof Y

Estimate ofY from regressionequation

Deviation,or error

Regression analysis in forecasting

Two applications of regressions analysis in forecasting Time-series data

Independent variable is time Dependent variable is the variable that you want to forecast (i.e. demand)

Data is not time-series Independent variable is a known variable that can be used to predict (i.e.

advertising dollars, customer population) Dependent variable is the variable that you want to forecast (i.e. demand)

Regression analysis is the same in both applications

Armand, Inc.: Regression Analysis

Armand, Inc. is a chain of Italian restaurants located in a five-state area.

The most successful locations have been near college campuses.

Prior to opening a new restaurant, management requires a forecast of the yearly sales revenues.Such an estimate is used

in planning the restaurant capacity, personnel requirements, and to see if the operations costs are smaller than the predicted

revenue.

Armand, Inc.Student population Annual sales

Restaurant (1000s) ($1000s)1 2 582 6 1053 8 884 8 1185 12 1176 16 1377 20 1578 20 1699 22 149

10 26 202

Armand, Inc.

Annual sales and Student Population

0 5 10 15 20 25 30

Student Population (1000s)

Armand, Inc.

Annual sales and Student Population

0 5 10 15 20 25 30

Student Population (1000s)

Go To Excel

Armand, Inc.

bXaY Intercept

Coefficient for Student Population

XY 560

Armand, Inc.

Forecast the Annual Sales if the student population is 20,000.

XY 560

Armand, Inc.

Forecast the Annual Sales if the student population is 20,000.

)20(560

XY Forecast is :

$160,000

Forecasting accuracy

“I think there is a world market for about FIVE computers.”

— Thomas J. Watson, chairman of IBM, 1943

Forecast accuracy

IBM 1994

$700 million inventory of OBSOLETE PCs that took 6

months to unload.

Reaction: too conservative when releasing the new

Aptiva home PCs. New models sold out before the

holiday season had begun.

Measuring the quality of forecasting

MAD – mean absolute deviation

MSE – mean square error

rorsForecastErMAD n

rorForecastErMSE n

Your Turn Demand for April-September is given. Determine the exponential smoothing forecasts for those

April. Forecast for Mar was 58 Demand for Mar was 60.

Determine the regression equation forecasts for those April. X is the number of months in the future (for April, X = 1)

Your TurnDemand Exponential

Smoothing alpha = 0.2

Regression Y = 54 + 3.9X

April 60

May 55

June 75

July 60

August 80

September 75

Calculate for APRIL:

Exponential smoothing forecast

Regression forecast

Forecast for Mar was 58Demand for Mar was 60.X is the number of months in the future (for April, X = 1)

Your Turn

Demand

Exponential Smoothing alpha = 0.20

April 60 58.4 1.6 57.9 2.1

August

September

Your Turn

Demand

Exponential Smoothing

alpha = 0.20Exp Smooth

abs(forecast error) Regression Y = 54 + 3.9X

Regressionabs(forecast error)

April 60 58.4 57.9

May 55 58.7 61.8

June 75 58.0 65.7

July 60 61.4 69.6

August 80 61.1 73.5

September 75 64.9 77.4

MAD = MAD =

Calculate abs(forecast error) for April

Your Turn

Demand

April 60 58.4 1.6 57.9 2.1

May 55 58.7 61.8

June 75 58.0 65.7

July 60 61.4 69.6

August 80 61.1 73.5

September 75 64.9 77.4

MAD = MAD =

Your Turn

Demand

April 60 58.4 1.6 57.9 2.1

May 55 58.7 3.7 61.8 6.8

June 75 58.0 17.0 65.7 9.3

July 60 61.4 1.4 69.6 9.6

August 80 61.1 18.9 73.5 6.5

September 75 64.9 10.1 77.4 2.4

MAD = MAD =

Calculate MAD for each.

Your Turn

Demand

April 60 58.4 1.6 57.9 2.1

May 55 58.7 3.7 61.8 6.8

June 75 58.0 17.0 65.7 9.3

July 60 61.4 1.4 69.6 9.6

August 80 61.1 18.9 73.5 6.5

September 75 64.9 10.1 77.4 2.4

MAD = 8.8 MAD = 6.12

Third Wave Research Group

- offers marketing software and databases

- Forecasts sales for specific

-Market areas

-Products

-segments

Tracking SignalsTracking Signals

Tracking SignalsTracking SignalsTracking signal =

+2.0 —

+1.5 —

+1.0 —

+0.5 —

–0.5 —

–1.0 —

–1.5 —| | | | |

0 5 10 15 20 25 Observation number

Control limit

+2.0 —

+1.5 —

+1.0 —

+0.5 —

–0.5 —

–1.0 —

–1.5 —| | | | |

Control limit

Out of control

MoMo FcstFcst ActAct ErrorError RSFERSFE AbsAbsErrorError

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

|Error||Error|

Tracking Signal Computation

MoMo ForcForc ActAct ErrorError RSFERSFE AbsAbsErrorError

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10

Error = Actual - Forecast = 90 - 100 = -10

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

6 100 140

-10-10 -10-10

RSFE = Errors = NA + (-10) = -10

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010

Abs Error = |Error| = |-10| = 10

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010

Cum |Error| = |Errors| = NA + 10 = 10

|Error||Error|

CumCum|Error||Error|

MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0

MAD = |Errors|/n = 10/1 = 10

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

TS = RSFE/MAD = -10/10 = -1

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

Error = Actual - Forecast = 95 - 100 = -5

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

-5-5 -15-15

RSFE = Errors = (-10) + (-5) = -15

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

-5-5 -15-15 55

Abs Error = |Error| = |-5| = 5

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

-5-5 -15-15 55 1515

Cum Error = |Errors| = 10 + 5 = 15

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

-5-5 -15-15 55 1515 7.57.5

MAD = |Errors|/n = 15/2 = 7.5

|Error||Error|

CumCum MADMAD TSTS

11 100100 9090

22 100100 9595

33 100100 115115

44 100100 100100

55 100100 125125

66 100100 140140

-10-10 -10-10 1010 1010 10.010.0 -1-1

-5-5 -15-15 55 1515 7.57.5 -2-2

|Error||Error|

TS = RSFE/MAD = -15/7.5 = -2

+2.0 —

+1.5 —

+1.0 —

+0.5 —

–0.5 —

–1.0 —

–1.5 —| | | | |

Control limit

Out of control

Chapter 4 Forecasting Production Planning Overview What is forecasting? Types of forecasts 7 steps of forecasting Qualitative forecasting.

time series

timetimeseries methodssimple

series of arithmetic

time horizon

forecasting modelsgather

forecasting methods

aggregated product forecasts

regular time periodsforecast

Documents

Sales Forecasting Introduction Sales forecasting is a...

chpt3 forecasting handout -...

Objective Techniques In Analog Seasonal Forecasts Objective....

Econometric Forecasting -...

Using Financial Forecasts to Advise Business - Method of...

Forecasting & Replenishment with IBM DIOS DIOS Demand...

Forecasting River Uruguay ﬂow using rainfall forecasts...

Demand Forecasting - DSpace@MIT:...

Econometric Forecasts Package: Short-Run Forecasting Models....

Forecasting Stock prices using Qualitative Data

(Production Forecasting and Decision Making) ·...

Forecasting February 26, 2007. Laws of Forecasting Three...

Agenda of Week VI. Forecasting and Design Forecasting...

Qualitative Sales Forecasting - SAGE - the natural home for

Weather Forecasting for Load Forecasts DTN/Meteorlogix.

Information, Forecasting and Corruption · Information,...