FORECASTING

Chapter Eighteen
McGraw-Hill/Irwin Copyright © 2014 by The McGraw-Hill Companies, Inc. All rights reserved.
Learning Objectives
 LO18–1: Understand how forecasting is essential
to supply chain planning.
 LO18–2: Evaluate demand using quantitative
forecasting models.
 LO18–3: Apply qualitative techniques to forecast
demand.
 LO18–4: Apply collaborative techniques to forecast
demand.

18-2
The Role of Forecasting
 Forecasting is a vital function and affects every significant
management decision.
 Finance and accounting use forecasts as the basis for budgeting
and cost control.
 Marketing relies on forecasts to make key decisions such as new
product planning and personnel compensation.
 Production uses forecasts to select suppliers; determine capacity
requirements; and drive decisions about purchasing, staffing, and
inventory.
 Different roles require different forecasting approaches.
 Decisions about overall directions require strategic forecasts.
 Tactical forecasts are used to guide day-to-day decisions.

18-3
 What’s Forecasted in the
Supply Chain?

•Demand, sales or requirements

•Purchase prices
•Replenishment and delivery
times

8-4 CR (2004) Prentice Hall, Inc.

Forecasting and Decoupling Point
 Decoupling point: Point at which inventory is stored,
which allows SC to operate independently
 The choice of the decoupling point in a SC is
strategic.
 Forecasting helps determine the level of
inventory needed at the decoupling points.
 The decision will be affected by the error produced
in the forecast and the type of product (easily
inventoried or easily perishable).

18-5
Types of Forecasting
 There are four basic types of forecasts.
1. Qualitative
2. Time series analysis (primary focus of this chapter)
3. Causal relationships
4. Simulation
 Time series analysis is based on the idea that
data relating to past demand can be used to
predict future demand.

18-6
Components of Demand

Average
demand for a Trend
period of time

Seasonal Cyclical
element elements

Random Autocorrelation
variation

18-7
 Typical Time Series Patterns:
Random
250

200
Sales

150
Actual sales
100 Average sales

50

0
0 5 10 15 20 25
Time

8-8 CR (2004) Prentice Hall, Inc.

 Typical Time Series Patterns:
Random with Trend
250

200
Sales

150

100
Actual sales
50 Average sales

0
0 5 10 15 20 25
Time

8-9 CR (2004) Prentice Hall, Inc.

 Typical Time Series Patterns:
Random with Trend & Seasonal

800
700
600
500
Sales

400
300
200 Actual sales
Trend in sales
100 Smoothed trend and seasonal sales
0
0 10 20 30 40
Time

8-10 CR (2004) Prentice Hall, Inc.

 Typical Time Series Patterns:
Lumpy
Sales

Time

8-11 CR (2004) Prentice Hall, Inc.

 Is Time Series Pattern
Forecastable?
Whether a time series can be reasonably
forecasted often depends on the time series’
degree of variability. Forecast a regular time
series, but use other techniques for lumpy ones.
How to tell the difference:
Rule
A time series is lumpy if

X  3
where
X  mean of the series
  standard deviation of series,
8-12 CR (2004) Prentice Hall, Inc.
Trends
 Identification of trend lines is a common starting
point when developing a forecast.
 Common trend types include linear, S-curve,
asymptotic, and exponential.

18-13
Time Series Analysis
 Using the past to predict the future
Short term – forecasting less than 3 months

• Used mainly for tactical decisions

Medium term – forecasting 3 months to 2 years

• Used to develop a strategy that will be implemented over the next

6 to 18 months (e.g., meeting demand)

Long term – forecasting greater than 2 years

• Useful for detecting general trends and identifying major turning

points

18-14
Model Selection
 Choosing an appropriate forecasting model
depends upon
1. Time horizon to be forecast
2. Data availability
3. Accuracy required
4. Size of forecasting budget
5. Availability of qualified personnel

18-15
 Forecasting Method Selection Guide
Forecasting Method Amount of Historical Data Data Pattern Forecast
Horizon
Simple moving average 6 to 12 months; weekly Stationary (i.e., no Short
data are often used trend or
seasonality)
Weighted moving 5 to 10 observations Stationary Short
average and simple needed to start
exponential smoothing
Exponential smoothing 5 to 10 observations Stationary and Short
with trend needed to start trend

Linear regression 10 to 20 observations Stationary, trend, Short to

and seasonality medium

18-16
 Simple Moving Average
 Forecast is the average of a fixed number of past
periods.
 Useful when demand is not growing or declining
rapidly and no seasonality is present.
 Removes some of the random fluctuation from the data.
 Selecting the period length is important.
 Longer periods provide more smoothing.
 Shorter periods react to trends more quickly.

18-17
Simple Moving Average Formula


18-18
Simple Moving Average – Example

18-19
 Weighted Moving Average
 The simple moving average formula implies equal
weighting for all periods.
 A weighted moving average allows unequal
weighting of prior time periods.
 The sum of the weights must be equal to one.
 Often,
more recent periods are given higher weights
than periods farther in the past.

18-20
 Example

Month 1 Month 2 Month 3 Month 4 Month 5

Demand 100 90 105 95 ?
Weight 10% 20% 30% 40%

 Forecast for Month 5=

 (0.4)(95)+(0.3)(105)+(0.20)(90)+(0.1)(100) = 97.5

21
Selecting Weights
 Experience and/or trial-and-error are the
simplest approaches.
 The recent past is often the best indicator of the
future, so weights are generally higher for more
recent data.
 If the data are seasonal, weights should reflect this
appropriately.

18-22
Exponential Smoothing
 A weighted average method that includes all past data in
the forecasting calculation
 More recent results weighted more heavily
 The most used of all forecasting techniques
 An integral part of computerized forecasting
 Well accepted for six reasons
1. Exponential models are surprisingly accurate.
2. Formulating an exponential model is relatively easy.
3. The user can understand how the model works.
4. Little computation is required to use the model.
5. Computer storage requirements are small.
6. Tests for accuracy are easy to compute.

18-23
Exponential Smoothing Model

18-24
 Exponential Smoothing Example
Week Demand Forecast
1 820 820
2 775 820
3 680 811
4 655 785
5 750 759
6 802 757
7 798 766
8 689 772
9 775 756
10 760

18-25
Trend-Corrected Exponential Smoothing
(Holt’s Model)

• Appropriate when the demand is assumed to

have a level and trend in the systematic
component of demand but no seasonality

Systematic component of demand = level + trend

Trend-Corrected Exponential Smoothing
(Holt’s Model)
 Obtain initial estimate of level and trend by
running a linear regression
Dt = a + bt
L0 = a, T0 = b
 In Period t, the forecast for future periods is
Ft+1 = Lt + Tt and Ft+n = Lt + nTt
 Revised estimates for Period t
Lt+1 = aDt+1 + (1 – a)(Lt + Tt)
Tt+1 = b(Lt+1 – Lt) + (1 – b)Tt
Trend-Corrected Exponential Smoothing
(Holt’s Model)
• Camera demand
D1 = 8,415, D2 = 8,732, D3 = 9,014,
D4 = 9,808, D5 = 10,413, D6 = 11,961
a = 0.1, b = 0.2
• Using regression analysis
L0 = 7,367 and T0 = 673
• Forecast for Period 1
F1 = L0 + T0 = 7,367 + 673 = 8,040
Trend-Corrected Exponential Smoothing
(Holt’s Model)
SUMMARY OUTPUT

Regression Statistics
Multiple R 0.95566483
R Square 0.91329526
Adjusted R Square 0.89161908
Standard Error 433.951402
Observations 6

ANOVA
df SS MS F Significance F
Regression 1 7934336 7934336 42.13358 0.002904838
Residual 4 753255.3 188313.8
Total 5 8687591

Standard
Coefficients Error t Stat P-value Lower 95% Upper 95% Lower 95.0% Upper 95.0%
Intercept 7,367 403.9868 18.23608 5.32E-05 6245.486245 8488.78042 6245.48624 8488.780422
X Variable 1 673 103.7342 6.491039 0.002905 385.3304729 961.355241 385.330473 961.3552414

L0 = 7,367, T0 = 673

29
Trend-Corrected Exponential Smoothing
(Holt’s Model)
• Revised estimate
L1 = aD1 + (1 – a)(L0 + T0)
= 0.1 x 8,415 + 0.9 x 8,040 = 8,078
T1 = b(L1 – L0) + (1 – b)T0
= 0.2 x (8,078 – 7,367) + 0.8 x 673
= 681
• With new L1, T1
F2 = L1 + T1 = 8,078 + 681 = 8,759
• Continuing
F7 = L6 + T6 = 11,399 + 673 = 12,074
Trend-Corrected Exponential Smoothing
(Holt’s Model)
Alpha = 0.1
Beta = 0.2

Period (t) Demand Lt Tt Lt+Tt Ft

0 7,367 673 8,040
1 8415 8,078 681 8,759 8,040
2 8,732 8,756 680 9,436 8,759
3 9,014 9,394 672 10,066 9,436
4 9,808 10,040 667 10,707 10,066
5 10,413 10,677 661 11,338 10,707
6 11,961 11,401 673 12,074 11,338
7 12,074

31
Choosing Alpha and Delta


18-32
Linear Regression Analysis
 Regression is used to identify the functional
relationship between two or more correlated
variables, usually from observed data.
 One variable (the dependent variable) is predicted
for given values of the other variable (the
independent variable).
 Linear regression is a special case that assumes the
relationship between the variables can be
explained with a straight line.

Y = a + bt

18-33
 Example 18.2 – Least Squares Method
The least squares method determines Quarter Sales Quarter Sales
the parameters a and b such that the 1 600 7 2,600
sum of the squared errors is 2 1,550 8 2,900
minimized – “least squares” 3 1,500 9 3,800
4 1,500 10 4,500
5 2,400 11 4,000
6 3,100 12 4,900

18-34
Example 18.2 – Calculations

1 600 600 1 360,000 801.3

2 1,550 3,100 4 2,402,500 1,160.9

3 1,500 4,500 9 2,250,000 1,520.5

4 1,500 6,000 16 2,250,000 1,880.1

5 2,400 12,000 25 5,760,000 2,239.7

6 3,100 18,600 36 9,610,000 2,599.4

7 2,600 18,200 49 6,760,000 2,959.0

8 2,900 23,200 64 8,410,000 3,318.6

The forecast is extended to periods 13-16
9 3,800 34,200 81 14,440,000 3,678.2

10 4,500 45,000 100 20,250,000 4,037.8

11 4,000 44,000 121 16,000,000 4,397.4

12 4,900 58,800 144 24,010,000 4,757.1

Sum 78 33,350 268,200 650 112,502,500

18-35
Regression with Excel
 Microsoft
Excel includes
Data
Analysis
tools, which
can perform
least squares
regression on
a data set.
Time Series Decomposition

 Chronologically ordered data are referred to as a

time series.
 A time series may contain one or many elements.
 Trend, seasonal, cyclical, autocorrelation, and
random
 Identifying these elements and separating the time
series data into these components is known as
decomposition.

18-37
Seasonal Variation
 Seasonal variation may be either additive or
multiplicative (shown here with a changing trend).

18-38
Determining Seasonal Factors :
Simple Proportions Example 18.3
 The seasonal factor (or index) is the ratio of the
amount sold during each season divided by the
average for all seasons.
Season Past Sales Average Seasonal
Sales for Factor
Each Season
Spring 200

Summer 350

Fall 300

Winter 150

Total 1000
18-39
Example 18.3 Continued

Expected Average Seasonal Next

Demand Sales for Factor Year’s
for Each Seasonal
Next Year Season Forecast
(1,100y4)
Spring 275 X 0.8 = 220
Summer 275 X 1.4 = 385
Fall 275 X 1.2 = 330
Winter 275 X 0.6 = 165
1100

18-40
Decomposition Using Least Squares
Regression
1. Decompose the time series into its components.
a. Find seasonal component/index.
b. Deseasonalize the demand (divide demand with
the seasonal index)
c. Find trend component.
2. Forecast future values of each component.
a. Project trend component into the future.
b. Multiply trend component by seasonal
component/index.

18-41
Decomposition – Steps 1 and 2
Using the data for periods 1-12, apply time series analysis
(decomposition, linear regression, trend estimate & seasonal
indices) to forecast for periods 13-16

18-42
 Decomposition – Steps 3 and 4
 Develop a least squares regression line for the
deseasonalized data.
 Project the regression line through the period of the forecast.

Regression Results:
Y = 555.0 + 342.2t

Forecast for
periods 13-16

18-43
 Decompostion – Step 5
 Create the final forecast by adjusting the regression
line by the seasonal factor.
Period Quarter Y from Regression Seasonal Factor Forecast
(F x Seasonal Factor
13 I 5,003.5 0.82 4,102.87
14 II 5,345.7 1.10 5,880.27
15 III 5,687.9 0.97 5,517.26
16 IV 6,030.1 1.12 6,753.71

18-44
Forecast Errors
 Forecast error is the difference between the forecast
value and what actually occurred.
 All forecasts contain some level of error.
 Sources of error
 Bias – when a consistent mistake is made
 Random – errors that are not explained by the model
being used
 Measures of error
 Mean absolute deviation (MAD)
 Mean absolute percent error (MAPE)
 Tracking signal

18-45
 Forecast Error Measurements
 Ideally, MAD will be zero  MAPE scales the forecast error to
(no forecasting error). the magnitude of demand.
 Larger values of MAD
indicate a less accurate
model.
 Tracking signal indicates whether
forecast errors are accumulating
over time (either positive or
negative errors).

18-46
Computing Forecast Error

18-47
Causal Relationship Forecasting
 Causal relationship forecasting uses independent
variables other than time to predict future demand.
 This independent variable must be a leading
indicator.
 Many apparently causal relationships are actually
just correlated events – care must be taken when
selecting causal variables.

18-48
Multiple Regression Techniques
 Often, more than one independent variable may be
a valid predictor of future demand.
 In this case, the forecast analyst may utilize multiple
regression.
 Analogous to linear regression analysis, but with
multiple independent variables.
 Multiple regression supported by statistical software
packages.

18-49
Qualitative Forecasting Techniques
 Generally used to take advantage of expert
knowledge.
 Useful when judgment is required, when products
are new, or if the firm has little experience in a
new market.
 Examples
 Market research
 Panel consensus
 Historical analogy
 Delphi method

18-50
 Collaborative Planning, Forecasting,
and Replenishment (CPFR)
 A web-based process used to coordinate the efforts of
a supply chain.
 Demand forecasting
 Production and purchasing

 Inventory replenishment

 Integrates all members of a supply chain –

manufacturers, distributors, and retailers.
 Depends upon the exchange of internal information
to provide a more reliable view of demand.

18-51
CPFR Steps

Creation of a Joint Development

front-end business of demand Sharing Inventory
partnership planning forecasts forecasts replenishment
agreement

18-52
Principles
 Forecasting is a fundamental step in any planning
process.
 Forecast effort should be proportional to the
magnitude of decisions being made.
 Web-based systems (CPFR) are growing in
importance and effectiveness.
 All forecasts have errors – understanding
and minimizing this error is the key to
effective forecasting processes.
18-53