Operations Management

Dr. TRAN QUYNH LE

Industrial Systems Engineering Department
Mechanical Engineering Faculty
Ho Chi Minh City University of Technology (HCMUT)–
VNUHCM
CHAPTER 3: FORECASTING

2
 LEARNING OUTCOME
After completing this chapter, you should be able to:

• List features common to all forecasts.

• List the elements of a good forecast.

• Outline the steps in the forecasting process.

• Describe four qualitative forecasting techniques.

• Use naive method, moving average forecast, weighted-average forecast, exponential

smoothing forecast, linear trend forecast, and trend-adjusted exponential smoothing
forecast.

• Construct control charts and use them to monitor forecast errors.

• Describe the key factors and trade-offs to consider when choosing a forecasting
technique.

3
 What is Forecasting?

• Forecasts are predictions, projections or estimates of future

events or conditions in the environment in which an enterprise
operates.

Or

• Forecasts are estimates of occurrence, timing or magnitude of

uncertain future events

4
 What is Forecasting?

• Process of predicting a future event

• Underlying basis of all business decisions

✓ Production
✓ Inventory
✓ Personnel
✓ Facilities
✓ ??
5
 What is Forecasting?

• The estimates can be

✓ Occurrence of an event
✓ Timing of happening
✓ Magnitude or volume
• The purpose of forecasting is to use the best available
information to guide future activities toward organizational
goals.

6
 What is Forecasting?
• Forecasting can be used to determine:

✓ Demand of product or service

✓ prices or costs

✓ Change of technology

✓ Competitors

✓ Interest rate

✓ Exchange rate

• Good forecasts enables managers to plan and budget for appropriate levels of personnel, raw
materials, capitals, inventory and a lot of other variables.

• Forecasting is the basis for all planning and budgeting efforts.

7
 Distinguishing Differences

• Medium/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 forecasting

• Short-term forecasts tend to be more accurate than longer-

term forecasts

8
Types of Forecasts

• Economic forecasts

✓ Address business cycle – inflation rate, money supply, etc.

• Technological forecasts

✓ Predict rate of technological progress

✓ Impacts development of new products
• Demand forecasts

✓ Predict sales of existing products and services

9
Strategic Importance of Forecasting

• Accounting. New product/process cost estimates, profit

projections, cash management.

• Finance. Equipment/equipment replacement needs, timing and

amount of funding/borrowing needs.

• Human resources. Hiring activities, including recruitment,

interviewing, and training; layoff planning, including
outplacement counseling

10
Strategic Importance of Forecasting

• Marketing. Pricing and promotion, e-business strategies, global

competition strategies.

• MIS. New/revised information systems, internet services.

• Operations. Schedules, capacity planning, work assignments

and workloads, inventory planning, make-or-buy decisions,
outsourcing, project management.

• Product/service design. Revision of current features, design of

new products or services

11
ELEMENTS OF A GOOD FORECAST

• The forecast should be timely.

• The forecast should be accurate, and the degree of accuracy
should be stated
• The forecast should be reliable; it should work consistently
• The forecast should be expressed in meaningful units
• The forecast should be in writing.
• The forecasting technique should be simple to understand and use.
• The forecast should be cost-effective: The benefits should outweigh
the costs.
12
 STEPS IN THE FORECASTING PROCESS

1. Determine the purpose of the forecast.

2. Establish a time horizon.

3. Obtain, clean, and analyze appropriate data.

4. Select a forecasting technique.

5. Make the forecast.

6. Monitor the forecast errors.

13
APPROACHES TO FORECASTING

• Judgmental forecasts: Forecasts that use subjective inputs

such as opinions from consumer surveys, sales staff,
managers, executives, and experts.

• Time-series forecasts: Forecasts that project patterns identified

in recent time-series observations.

• Associative model: Forecasting technique that uses

explanatory variables to predict future demand.

14
QUALITATIVE FORECASTS

• Executive Opinions: Pool opinions of high-level experts,

sometimes augment by statistical models

✓ Involves small group of high-level experts and managers

✓ Group estimates demand by working together
✓ Combines managerial experience with statistical models
✓ Relatively quick
✓ ‘Group-think’ disadvantage
15
QUALITATIVE FORECASTS

• Sales force Opinions: Estimates from individual salespersons

are reviewed for reasonableness, then aggregated

✓ Each salesperson projects his or her sales

✓ Combined at district and national levels
✓ Sales reps know customers’ wants
✓ Tends to be overly optimistic

16
QUALITATIVE FORECASTS

• Consumer Surveys: Ask the customer

✓ Ask customers about purchasing plans

✓ What consumers say, and what they actually do are often
different

✓ Sometimes difficult to answer

17
QUALITATIVE FORECASTS

• Delphi method: An iterative process in which managers and staff

complete a series of questionnaires, each developed from the
previous one, to achieve a consensus forecast.

✓ Iterative group process, continues until consensus is reached

✓ 3 types of participants
➢ Decision makers
➢ Staff
➢ Respondents
18
FORECASTS BASED ON TIME-SERIES DATA

• Time Series Components:

✓ Trend A long-term upward or downward movement in data.
✓ Seasonality Short-term regular variations related to the
calendar or time of day.
✓ Cycle Wavelike variations lasting more than one year.
✓ Irregular variation Caused by unusual circumstances, not
reflective of typical behavior.
✓ Random variations Residual variations after all other
behaviors are accounted for
19
FORECASTS BASED ON TIME-SERIES DATA

The graph shows the average, trend, and

Seasonal variation seasonal components and randomness
around the smoothed demand curve.
x
x x Linear
x x
x Trend
x x
x
x x x
Sales

xx x x
x xx x x
x
x
x x x x x x
x x x x x x
x x x
x xxxxx
x
x x

1 2 3 4
Year
20
FORECASTS BASED ON TIME-SERIES DATA

• Naive approach

• Moving averages

• Exponential smoothing

• Trend projection

• Linear regression

• Time-Series Models

• Associative Model
21
Naive Methods

• A forecast for any period that equals the previous period’s

actual value

• e.g., If January sales were 68, then February sales will be 68

• Sometimes cost effective and efficient

• Can be good starting point

22
Naive Methods
Period, t Demand Forecast
1 12
2 13
3 20
4 23
5 25
6 16
7 14
8 25
9 23
10 22

23
Naive Methods
Period, t Demand Forecast
1 12
2 13 12
3 20 13
4 23 20
5 25 23
6 16 25
7 14 16
8 25 14
9 23 25
10 22 23

24
Moving Average
• MA works best when there are no trends or seasonality.
• A moving average of order N or MA(N) is an average of the most
recent N observations.

Ft = (Dt-N + Dt-N+1 + … + Dt-2 + Dt-1)/N

Ft = Forecast demand in period

Dt-N= Actual demand in period t-N
N = The number of periods to use in computing average

25
Moving Average
• N can be from 2-12 periods, with 3-4 periods being
common.

• Small values of N put more weight on current data.

• Large values of N should be used for demand series that

are stable.

• Small values of N are for those that are susceptible to

changes in the underlying average.

26
Moving Average

Period Month Dt Ft et
1 Jan 90
2 Feb 106
3 Mar 152
4 Apr 244 116.00 128.00
5 May 302 167.33 134.67
6 Jun 274 232.67 41.33
7 Jul 162 273.33 -111.33
8 Aug 194 246.00 -52.00
: : : : :
20 Aug 182 279.00 -97.00
21 Sep 323 223.33 99.67
22 Oct 400 227.67 172.33
23 Nov 252 301.67 -49.67
24 Dec 179 325.00 -146.00

27
 Moving Average

Period Month Dt Ft et
1 Jan 90
2 Feb 106 F4 = (D1 + D2 + D3)/3
3 Mar 152 = (90+106+152)/3
4 Apr 244 116.00 128.00
= 116.00
5 May 302 167.33 134.67
6 Jun 274 232.67 41.33
7 Jul 162 273.33 -111.33 F5 = (D2 + D3 + D4)/3
8 Aug 194 246.00 -52.00 = (106+152+244)/3
: : : : : = 167.33
20 Aug 182 279.00 -97.00
21 Sep 323 223.33 99.67
22 Oct 400 227.67 172.33
23 Nov 252 301.67 -49.67
24 Dec 179 325.00 -146.00

28
Moving Average
• MA lags behind any trends.

• Lag Effect: The forecast is consistently later than (behind) the

original data.

MA(3) vs. MA(6)

600

500

400
Demand

300

200

100

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Actual MA(3) MA(6)

29
Weighted Moving Average (WMA)
• Most recent data are more indicative of the future than are older
data. Thus, more weight may be given to the more recent data.

• Weighting data may reduce some of the lagging found in simple

MA.

• Weights are the subjective evaluation of the forecaster of the

importance of more recent data and older data in making a
forecast.

30
Weighted Moving Average (WMA)

Ft = (Wt-NDt-N + … + Wt-2Dt-2 + Wt-1Dt-1) / W

Ft = Forecast demand in period t

Dt-N = Actual demand in period t-N

Wt-N = Weight of demand in period t-N

N = The number of periods to use in computing average

31
Weighted Moving Average (WMA)
Period Month Dt Ft et
1 Jan 90
2 Feb 106
3 Mar 152
4 Apr 244 116.00 128.00
5 May 302 167.33 134.67
6 Jun 274 232.67 41.33
7 Jul 162 273.33 -111.33
8 Aug 194 246.00 -52.00
: : : : :
20 Aug 182 279.00 -97.00
21 Sep 323 223.33 99.67
22 Oct 400 227.67 172.33
23 Nov 252 301.67 -49.67
24 Dec 179 325.00 -146.00

32
Weighted Moving Average (WMA)
WMA with weights of 2,3 and 4:
F4 = (2D1 + 3D2 + 4D3)/(2+3+4)
= (2/9)90+(3/9)106+(4/9)152
= 122.89

F5 = (2D2 + 3D3 + 4D4)/(2+3+4)

= (2/9)106+(3/9)152+(4/9)244
= 182.67

WMA with weights of 1,5 and 9:

F4 = (1D1 + 5D2 + 9D3)/(1+5+9)
= (1/15)90+(5/15)106+(9/15)152
= 132.53

F5 = (1D2 + 5D3 + 9D4)/(1+5+9)

= (1/15)106+(5/15)152+(9/15)244
= 204.13

33
Weighted Moving Average (WMA)

MA(3) vs. WMA(3)

600

500

400
Demand

300

200

100

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Actual MA(3) WMA(3)-(2,3,4) WMA(3)-(1,5,9)

34
Simple Exponential Smoothing (SES)
• First-Order Exponential Smoothing

• Exponentially Weighted Moving Average

• A forecasting technique based on the forecasting errors

• The current forecast is the weighted average of the last

forecast and the current value of demand.

• Offers the method that is equivalent to weighted moving

average but requires fewer data (i.e., the last forecast, the
last actual demand and ) and calculations.

35
Simple Exponential Smoothing (SES)

Ft = Ft-1 + (Dt-1 - Ft-1)

Ft = Dt-1 + (1 - )Ft-1

Ft = Forecast demand in period t

Dt-1 = Actual demand in period t-1

Ft-1 = Forecast demand in period t-1

 = Weight; 0    1

36
Simple Exponential Smoothing (SES)
Period Month Dt Ft et
1 Jan 90
2 Feb 106
3 Mar 152
4 Apr 244 116.00 128.00
5 May 302 167.33 134.67
6 Jun 274 232.67 41.33
 = 0.1
7 Jul 162 273.33 -111.33
8 Aug 194 246.00 -52.00
: : : : :
20 Aug 182 279.00 -97.00
21 Sep 323 223.33 99.67
22 Oct 400 227.67 172.33
23 Nov 252 301.67 -49.67
24 Dec 179 325.00 -146.00

37
Simple Exponential Smoothing (SES)
SES ( =0.1)
Period Month Dt Ft et
1 Jan 90 90.00 0.00
2 Feb 106 90.00 16.00
3 Mar 152 91.60 60.40
4 Apr 244 97.64 146.36
5 May 302 112.28 189.72  = 0.1
6 Jun 274 131.25 142.75 F1 = D1
7 Jul 162 145.52 16.48 F2 = 0.1D1+0.9F1
8 Aug 194 147.17 46.83 = 0.1(90) + 0.9(90) = 90
: : : : : F3 = 0.1D2+0.9F2
20 Aug 182 203.61 -21.61 = 0.1(106) + 0.9(90) = 91.6
21 Sep 323 201.45 121.55
22 Oct 400 213.61 186.39
23 Nov 252 232.25 19.75
24 Dec 179 234.22 -55.22

38
Simple Exponential Smoothing (SES)
SES: Ft = Dt-1 + (1 - )Ft-1

Ft-1 = Dt-2 + (1 - )Ft-2 substitute in (1)

Ft = Dt-1 + (1 - )Dt-2 + (1- )2 Ft-2

Thus, an infinite expansion for Ft:


Ft =   (1 −  ) i Dt −i −1
i =1
• Period t-1: Weight =  = 0.1

Period t-2: Weight = (1 - ) = 0.09

Period t-3: Weight = (1 - ) 2 = 0.081

Period t-n: Weight = (1 - )n-1

39
Simple Exponential Smoothing (SES)

• Assume that initial forecast is equal to initial

demand or average of the first few demands.

• Large  places weight on the most recent data &

tracking the series with trend better (but lagging
still occurs)

• Typically,  = [0.1, 0.3]

• If  > 0.5 is required, more sophisticated model

may be needed because of a significant trend or
seasonal influence in the series.
40
Simple Exponential Smoothing (SES)

SES (alpha=0.1) vs. SES (alpha=0.7)

600

500

400
Demand

300

200

100

0
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48

Actual SES (0.1) SES (0.7)

41
Comparison of SES & MA
Similarity

1. Both methods are derived with the assumption that the underlying
demand is stationary.

2. Both methods depend on the specification of a single parameter: N for

MA and  for SES.
Small N & large  (i.e., more weights) may be more responsive to
changes but will lead to forecast errors with high variance.

3. Both methods will lag behind a trend if it exists.

4. When  = 2/(N+1), both methods have the same distribution of

forecast error, i.e., both should have roughly the same level of
accuracy.

42
Comparison of SES & MA (Cont.)

Differences

1. SES is a weighted average of all past data points (as long as  < 1),
while MA is a weighted average of the last N periods of data. An
outlier is washed out for MA but remains forever in SES.

SES: An infinite expansion for Ft:


Ft =   (1 −  ) i Dt −i −1
i =1
2. To use MA, we must save all N past data points, while to use SES, we
need to save only the last forecast.
This is the reason for SES popularity in practice.

43
Linear trend forecast

Linear trend equation Ft = a + bt, used to develop forecasts when

trend is present.

44
Linear trend forecast

45
Linear trend forecast

46
Trend-adjusted exponential smoothing

Variation of exponential smoothing used when a time series

exhibits a linear trend.

47
Trend-adjusted exponential smoothing

48
Trend-adjusted exponential smoothing

49
Techniques for Seasonality

Seasonal variations Regularly repeating movements in series

values that can be tied to recurring events.

The multiplicative seasonal model can adjust trend data for

seasonal variations in demand

50
Seasonal Variations In Data

Steps in the process:

• Find average historical demand for each season

• Compute the average demand over all seasons

• Compute a seasonal index for each season

• Estimate next year’s total demand

• Divide this estimate of total demand by the number of seasons,

then multiply it by the seasonal index for that season

51
Seasonal Variations In Data

52
Seasonal Variations In Data

53
Seasonal Variations In Data

54
Seasonal Variations In Data

55
ASSOCIATIVE FORECASTING TECHNIQUES

• Used when changes in one or more independent variables can

be used to predict the changes in the dependent variable

• Most common technique is linear regression analysis

• We apply this technique just as we did in the time series

example

56
Evaluating Forecasts
• Forecast Error:
The difference between the actual demand & the forecast

ei = Di – Fi

ei = Forecast error at period i

Di = Actual demand at period i

Fi = Forecast at period i calculated in an earlier period.

 (e − e)
n 2

• Standard Deviation of Forecast Errors i

= i =1

n −1
57
Evaluating Forecasts (Cont.)
• Mean Absolute Deviation
(MAD) n n

e D −F i i i

MAD = i =1
= i =1

n n
• The most common measures of forecast error.
• The estimate of random error component.
• The MAD is often the preferred method of measuring the
forecast error because it does not require squaring.
• When forecast errors are normally distributed, as is
generally assumed, e  1.25(MAD)
58
Evaluating Forecasts (Cont.)
• Mean Squared Error (MSE) n n

 ei2  i i
( D − F ) 2

MSE = i =1
= i =1

n n
• Indicator of which of a set of forecasting techniques
produces the more usable results.

• The MSE is similar to the variance of a random sample.

• Root Mean Squared Error (RMSE) can also be used to

measure error in the same unit.

• RMSE is also the estimate of random error component.

59
Evaluating Forecasts (Cont.)
• Mean Absolute Percentage error, MAPE:

e i Di
MAPE = i =1
 100
n

• Not dependent on the magnitude of the values of demand.

60
Evaluating Forecasts (Cont.)

61
Evaluating Forecasts (Cont.)

62
Evaluating Forecasts (Cont.)

63