Operations Management

Forecasting
Chapter 4

4-1
 Examples
 Predict the next number in the pattern:

a) 3.7, 3.7, 3.7, 3.7, 3.7, ?

b) 2.5, 4.5, 6.5, 8.5, 10.5, ?

c) 5.0, 7.5, 6.0, 4.5, 7.0, 9.5, 8.0, 6.5, ?

4-2
 Examples
 Predict the next number in the pattern:

a) 3.7, 3.7, 3.7, 3.7, 3.7, y = 3.7

b) 2.5, 4.5, 6.5, 8.5, 10.5, y = 0.5 + 2x

c) 5.0, 7.5, 6.0, 4.5, 7.0, 9.5, 8.0, 6.5,

y = 4.5 + 0.5x + ci
c1 = 0; c2 = 2; c3 = 0; c4 = -2; etc
4-3
 Outline
 What is Forecasting?
 Time horizons.
 Life cycle.
 Types of Forecasts.
 Eight Steps in the Forecasting System.
 Forecasting Approaches:
 Overview of Qualitative Methods.
 Overview of Quantitative Methods.

4-4
 Outline - Continued
 Time-Series Forecasting:
 Moving Averages.
 Exponential Smoothing.
 Trend Projection.

 Associative Forecasting Methods: Regression and

Correlation Analysis.
 Monitoring and Controlling Forecasts.
 Forecasting in the Service Sector.

4-5
 What is Forecasting?
 Art and science of
predicting future events. Sales will
be $200
 Underlying basis of Million!
all business decisions.
 Production & Inventory.
 Personnel & Facilities.

 Focus on forecasting
demand.

4-6
 Types of Forecasts by Time Horizon
Short-range forecast: Usually < 3 months.
 Job scheduling, worker assignments.

Medium-range forecast: 3 months to 3 years.

 Sales & production planning, budgeting.

Long-range forecast: > 3 years.

 New product planning, facility location.

4-7
 Short- vs. Long-term Forecasting
 Medium & Long range forecasts:
 Long range for design of system.
 Deal with comprehensive issues.
 Support management decisions regarding planning.

 Short-term forecasts:
 To plan detailed use of system.
 Usually use quantitative techniques.
 More accurate than longer-term forecasts.

4-8
 Influence of Product Life Cycle
Stages of introduction and growth require longer
forecasts than maturity and decline.
Forecasts useful in projecting:
 staffing levels,
 inventory levels, and
 factory capacity (expansion and contraction),

as product passes through life cycle stages.

4-9
Forecasting During the Life Cycle
Introduction Growth Maturity Decline

Hard to forecast. Forecasting Easier to forecast. Hard to forecast,

critical, both for Use quantitative but forecasting is
Need long-range less important.
future magnitude models.
forecasts.
and growth rate.
Often use
Long-range
qualitative models.
forecasts still
important.

Sales

Time
4-10
 Eight Steps in Forecasting
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.
Monitor forecasts and adjust when needed.

4-11
 Realities of Forecasting
Assumes future will be like the past (causal factors
will be the same).

Forecasts are imperfect.

Forecasts for groups of product are more accurate

than forecasts for individual products.

Accuracy decreases with length of forecast.

4-12
 Forecasting Approaches
Qualitative Methods Quantitative Methods
 Used when little data or time  Used when situation is
exist. ‘stable’ & historical data exist.
 New products & technology.  Existing products & current
 Long time horizon. technology.
 Major changes expected.  No significant changes
expected.
 Involves intuition, experience.
 Example: forecasting for e-  Involves mathematical
commerce sales. techniques.
 Example: forecasting sales of
color televisions.

4-13
 Overview of Qualitative Methods
Jury of executive opinion.
 Combine opinions from executives.
Sales force composite.
 Aggregate estimates from salespersons.
Delphi method.
 Query experts interatively.
Consumer market survey.
 Survey current and potential customers.

4-14
 Jury of Executive Opinion
 Seek opinions/estimates from small group
of high-level managers working together.
 Combines managerial experience with
statistical models.

+ Relatively quick.
- ‘Group-think’.
- Leader may dominate.

4-15
 Sales Force Composite
 Each salesperson projects their
sales.
 Aggregate projections at district & Sales
national levels.

+ Sales rep’s know customers.

- Must not reward inaccurate
forecasts.
 May over- or under-forecast to acquire
more resources.

4-16
 Delphi Method
 Iterative group process.
 3 types of people: Decision Makers
 Decision makers. (Make forecast)
 Staff. Staff
 Respondents. (Administer)

+ Reduces ‘group-think’.
- Takes time.
Respondents
(Provide input to decision makers)

4-17
 Consumer Market Survey
How many hours will
 Ask customers about you use the Internet
purchasing plans. next week?

+ Relatively simple.
- What consumers say, and
what they actually do are
often different.

4-18
Quantitative Forecasting Methods

Quantitative
Forecasting

Time Series Associative

Models Models

Moving Exponential Trend Linear

Average Smoothing Projection Regression

4-19
 What is a Time Series?
 Set of evenly spaced numerical data.
 From observing response variable at regular time periods.
 Forecast based only on past values.
 Assumes that factors influencing past will continue
influence in future.
 Example:
Year: 1 2 3 4 5
Sales: 78.763.5 89.7 93.292.1

4-20
 Time Series Components

Trend Cyclical

Seasonal Random

4-21
Demand for product or service
Product Demand over 4 Years

Year Year Year Year

1 2 3 4
4-22
 Product Demand over 4 Years
Trend component
Seasonal peaks
Demand for product or service

Cyclic
component
Actual demand
Random line
variation
Year Year Year Year
1 2 3 4
4-23
 Trend Component
 Persistent, overall upward or downward pattern.
 Due to population, technology etc.
 Several years duration.

Time

4-24
 Seasonal Component
 Regular pattern of up & down fluctuations.
 Due to weather, customs etc.
 Occurs within 1 year.
 Quarterly, monthly, weekly, etc.
Summer
Demand

Time
4-25
 Cyclical Component
 Repeating up & down movements.
 Due to interactions of factors influencing economy.
 Usually 2-10 years duration.

Cycle
Demand

Year
4-26
 Random Component

 Erratic, unsystematic, ‘residual’ fluctuations.

 Due to random variation or unforeseen events.
 Union strike
 Tornado
 Short duration & non-repeating.

4-27
 General Time Series Models

 Any value in a time series is a combination of

the trend, seasonal, cyclic, and random
components.

 Multiplicative model: Yi = Ti · Si · Ci · Ri

 Additive model: Yi = Ti + Si + Ci + Ri

4-28
 Naive Approach
 Demand in next period is the same
as demand in most recent period.
 e.g., If May sales were 48, then June
sales will be 48.

 Sometimes cost effective & efficient.

 Usually not good.

4-29
 Moving Average Method
 MA is a series of arithmetic means.
 Used if little or no trend.
 Used often for smoothing.

MA   Demand in previous n periods

n

4-30
 Moving Average Example
You’re manager of a museum store that sells
historical replicas. You want to forecast sales (in
thousands) for months 4 and 5 using a 3-period
moving average.

Month 1 4
Month 2 6
Month 3 5
Month 4 ?
Month 5 ?
4-31
 Moving Average Forecast
Month Response Moving Moving
Yi Total Average
(n=3) (n=3)
1 4 NA NA
2 6 NA NA
3 5 NA NA
4 ? 4+6+5=15 15/3=5
5 ?
6 ?

4-32
 Moving Average Graph

Sales
8
Actual
6 Forecast
44
22
95
1 96
2 97
3 984 99
5 00
6
Month
4-33
 Actual Demand for Month 4 = 3
Month Response Moving Moving
Yi Total Average
(n=3) (n=3)
1 4 NA NA
2 6 NA NA
3 5 NA NA
4 3 4+6+5=15 15/3 = 5
5 ?
6 ?

4-34
 Moving Average Graph

Sales
8
Actual
6 Forecast
44
22
95
1 96
2 97
3 984 99
5 00
6
Month
4-35
 Moving Average Forecast
Month Response Moving Moving
Yi Total Average
(n=3) (n=3)
1 4 NA NA
2 6 NA NA
3 5 NA NA
4 3 15 5
5 7 6+5+3=14 14/3=4.667
6 ?

4-36
 Moving Average Graph

Sales
8
Actual
6 Forecast
44
22
95
1 96
2 97
3 984 99
5 00
6
Month
4-37
 Actual Demand for Month 5 = 7
Month Response Moving Moving
Yi Total Average
(n=3) (n=3)
1 4 NA NA
2 6 NA NA
3 5 NA NA
4 3 15 5
5 7 6+5+3=14 14/3=4.667
6 ?

4-38
 Moving Average Graph

Sales
8
Actual
6

44 Forecast
22
95
1 96
2 97
3 984 99
5 00
6
Month
4-39
 Moving Average Forecasts
Month Response Moving Moving
Yi Total Average
(n=3) (n=3)
1 4 NA NA
2 6 NA NA
3 5 NA NA
4 3 4+6+5=15 15/3=5.0
5 7 6+5+3=14 14/3=4.667
6 ? 5+3+7=15 15/3=5.0

4-40
 Moving Average Graph

Sales
8
Actual
6

44
Forecast
22

195 296 397 498 599 600

Month
4-41
 Weighted Moving Average Method
 Gives more emphasis to recent data.
 Weights decrease for older data.
 Weights sum to 1.0.
 May be based on intuition.
 Sum of digits weights: numerators are consecutive.
 3/6, 2/6, 1/6
 4/10, 3/10, 2/10, 1/10

Σ [(Weight for period n) (Demand in period n)]

WMA =
ΣWeights
4-42
Weighted Moving Average: 3/6, 2/6,
1/6
Month Response Weighted
Yi Moving
Average
1 4 NA
2 6 NA
3 5 NA
4 ? 31/6 = 5.167
5 ?
6 ?
4-43
Weighted Moving Average: 3/6, 2/6,
1/6
Month Response Weighted
Yi Moving
Average
1 4 NA
2 6 NA
3 5 NA
4 3 31/6 = 5.167
5 7 25/6 = 4.167
6 ? 32/6 = 5.333
4-44
 Moving Average Methods
 Increasing n makes forecast:
 Less sensitive to changes.
 Less sensitive to recent data.
 Weights control emphasis on recent data.
 Do not forecast trend well.
 Require historical data.

4-45
 Moving Average Graph

Actual

Demand

Time
4-46
 Moving Average Graph

Actual
Large n

Small n
Demand

Time
4-47
 Weighted Moving Average Graph
Small weight on
Actual recent data

Large weight
Demand on recent data

Time
4-48
 Exponential Smoothing Method
Form of weighted moving average.
 Weights decline exponentially.
 Most recent data weighted most.
Requires smoothing constant ().
 Usually ranges from 0.05 to 0.5
 Should be chosen to give good forecast.
Involves little record keeping of past data.

4-49
 Exponential Smoothing Equation

 Ft = Ft-1 + (At-1 - Ft-1)

 Ft = Forecast value for time t
 At-1 = Actual value at time t-1
 = Smoothing constant

 Need initial forecast Ft-1 to start.

 Could be given or use moving average.

4-50
 Exponential Smoothing Example
You want to forecast product demand using exponential
smoothing with  = .10. Suppose in the most recent
month (month 6) the forecast was 175 and the actual
demand was 180.
Month 6 180
Month 7 ?
Month 8 ?
Month 9 ?
Month 10 ?
4-51
 Exponential Smoothing - Month 7
Ft = Ft-1 + α (At-1 - Ft-1)
Forecast, F t
Month Actual
α(  = .10)
6 180 175.00 (Given)
7 ? 175.00 + .10(180 - 175.00) = 175.50
8 ?
9 ?
10 ?
11 ?

4-52
 Exponential Smoothing - Month 8
Ft = Ft-1 + α (At-1 - Ft-1)
Forecast, F t
Month Actual
( α = .10)
6 180 175.00 (Given)
7 168 175.00 + .10(180 - 175.00) = 175.50
8 ? 175.50 + .10(168 - 175.50) = 174.75
9 ?
10 ?
11 ?

4-53
 Exponential Smoothing Solution
Ft = Ft-1 + α (At-1 - Ft-1)
Forecast, F t
Month Actual
(α = .10)
6 180 175.00 (Given)
7 168 175.00 + .10(180 - 175.00) = 175.50
8 159 175.50 + .10(168 - 175.50) = 174.75
9 ? 174.75 + .10(159 - 174.75) = 173.18
10 ?
11 ?

4-54
 Exponential Smoothing Solution
Ft = Ft-1 + α (At-1 - Ft-1)
Forecast, F t
Month Actual
(α = .10)
6 180 175.00 (Given)
7 168 175.00 + .10(180 - 175.00) = 175.50
8 159 175.50 + .10(168 - 175.50) = 174.75
9 175 174.75 + .10(159 - 174.75) = 173.18
10 190 173.18 + .10(175 - 173.18) = 173.36
11 ? 173.36 + .10(190 - 173.36) = 175.02

4-55
Exponential Smoothing Graph

Sales
190 Actual
180
170 Forecast
160
150
140
6 7 8 9 10 11
Month
4-56
 Exponential Smoothing Methods
 Increasing α makes forecast:
 More sensitive to changes.
 More sensitive to recent data.

 α controls emphasis on recent data.

Do not forecast trend well.

 Trend adjusted exponential smoothing - p. 90-93

4-57
 Exponential Smoothing Graph

Actual

Demand

Time
4-58
 Exponential Smoothing Graph

Actual
Small α

Large α
Demand

Time
4-59
 Forecast Effects of
Smoothing Constant 

Ft =  At - 1 + (1- )At - 2 + (1- )2At - 3 + ...

Weights
= Prior Period 2 periods ago 3 periods ago
 (1 - ) (1 - )2

= 0.10 10% 9% 8.1%

= 0.90 90% 9% 0.9%

4-60
 Choosing  - Comparing Forecasts
A good method has a small error.
Choose  to produce a small error.

 Error = Demand - Forecast

Error > 0 if forecast is too low
Error < 0 if forecast is too high

MAD = Mean Absolute Deviation: Average of absolute

values of errors.
MSE = Mean Squared Error: Average of squared errors.
MAPE = Mean Absolute Percentage Error: Average of
absolute value of percentage errors.
4-61
 Forecast Error Equations
Mean Absolute Deviation (MAD)
n
| y i  yˆ i | | forecast errors |
MAD  i1 
n n

Mean Squared Error (MSE)

n
 (y i  yˆ i )2 2
MSE  i1

 forecast errors
n n

4-62
 Forecast Error Equations

Mean Absolute Percentage Error (MAPE)

| y i  yˆ i |
n
| forecast errors |
 
i1 y i Actual
MAPE  
n n

4-63
 Forecast Error Example
Actual F1 F1 error F2 F2 error
20 19 1 18 2
10 15 -5 13 -3
24 22 2 21 3
20 21 -1 18 2

MAD F1 = 9/4 = 2.25 F2 = 10/4 = 2.5

MSE F1 = 31/4 = 7.75 F2 = 26/4 = 6.5

MAPE F1 = 0.171 = 17.1% F2 = 0.156 = 15.6%

4-64
 Which Forecast is Best?

MAD F1 = 9/4 = 2.25 F2 = 10/4 = 2.5

MSE F1 = 31/4 = 7.75 F2 = 26/4 = 6.5

MAPE F1 = 0.171 = 17.1% F2 = 0.156 = 15.6%

4-65