2

Forecasting

McGraw-Hill/Irwin Copyright © 2007 by The McGraw-Hill Companies, Inc. All rights reserved.
 Learning Objectives
 List the elements of a good forecast.
 Outline the steps in the forecasting
process.
 Describe at least three qualitative
forecasting techniques and the advantages
and disadvantages of each.
 Compare and contrast qualitative and
quantitative approaches to forecasting.

3-2
 Learning Objectives
 Briefly describe averaging techniques,
trend and seasonal techniques, and
regression analysis, and solve typical
problems.
 Describe two measures of forecast
accuracy.
 Describe two ways of evaluating and
controlling forecasts.
 Identify the major factors to consider when
choosing a forecasting technique.
3-3
 How the car buyer think ?
 He wants it as soon as possible
 He does not have to wait any time for delivery
 If the dealer is not ready to deliver at once the
buyer look elsewhere
Hence

 It is important for a dealer to anticipate buyer

wants and to have those models, with the
necessary options, in stock
3-4
 How the car buyer think ?
The dealer who can correctly
forecast buyer wants, and who have
those cars available, is going to be
much more successful than a
competitor who guesses instead of
forecasting

 He usually
 Guesses wrong !!!!!!
 Gets stuck with cars customers don’t want !!!!!!!
 So, how does the dealer know how many cars of each type to stock ?
 The answer is :
The dealer doesn’t know for sure, but by analyzing previous buying
patterns, and perhaps making allowances for current conditions, the
dealer can come up with a reasonable approximation of what buyers will
want
3-5
 Planning and Uncertainties
 The manager is a planner
 Uncertainty make planning difficult
 Forecasting reduce uncertainty
Forecasting enable managers
to develop more meaningful
plans

so

What is forecasting ?
3-6
 What is forecasting ?
FORECASTING IS:
 A statement about the future value of a
variable of interest such as demand.
 Forecasting is used to make informed
decisions.
 Long-range (covering several years) that will
have long-term consequences on a country,
organization, town, city ……. )
 Short-range ( several weeks, days,…)
 People make and use forecasts all the time
both in their jobs and in everyday life
3-7
 Forecasts
 Forecasts affect decisions and activities
throughout an organization
 Accounting, finance
 Human resources
 Marketing
 MIS
 Operations
 Product / service design

3-8
 Uses of Forecasts
1. Help managers plan the system
( long term )
 Types of products to offer
 Facilities and equipment to have, to locate
2. Help managers plan the use of the system
( short term )
 Planning inventory
 Work force level
 Planning purchasing and production
 Budgeting and scheduling
3-9
 Uses of Forecasts

Cost/profit estimates Accounting

Cash flow and funding Finance

Hiring/recruiting/training Human Resources

Pricing, promotion, strategy Marketing

IT/IS systems, services MIS

Schedules, MRP, workloads Operations

New products and services Product/service design

3-10
 Features of Forecasts
 Assumes causal system
past ==> future
 Forecasts rarely perfect because of
randomness
 Forecasts more accurate for
groups of items vs. individual items
I see that you will
 Forecast accuracy decreases get an A this semester.

as time horizon increases

3-11
Elements of a Good Forecast

y
el
i m
T

t
le ate ffc
b r t -e
elia c cu
o s
R A C
e
f ul n u s
n g tt e to
ni ri y
a s
M
e W Ea

3-12
Steps in the Forecasting Process

“The forecast”

Step 6 Monitor the forecast

Step 5 Make the forecast
Step 4 Obtain, clean and analyze data
Step 3 Select a forecasting technique
Step 2 Establish a time horizon
Step 1 Determine purpose of forecast

3-13
 Approaches to forecasting

 Qualitative:
 Consists of subjective inputs
 Includes soft information (human factors,
personal opinion)
 Quantitative:
 Consists of objective inputs
 Involve hard data (projection of historical
data, analyzing objectives)

3-14
Types of forecasting techniques

 Judgmental - uses subjective inputs

obtained from various resources
 Time series - uses historical data
assuming the future will be like the past
 Associative models - uses explanatory
variables to predict the future

3-15
 Judgmental Forecasts
 Executive opinions
 Sales force opinions
 Consumer surveys
 Outside opinion
 Delphi method
 Opinions of managers and staff
 Achieves a consensus forecast

3-16
 Executive opinions
.Est. * prob probability estimation

50 40% 125 Production

manager

56 35% 160 marketing

manager

25 25% 100 Financial

manager

131 General
manager

3-17
 Delphi method
Session 4 Session 3 Session 2 Session 1

4 4 4.5 5 Expert 1

3.5 3.5 3.5 3 Expert 2

4 4.5 4 4.5 Expert 3

4 4.5 5 6 Expert 4

3-18
 Time Series Forecasts
Plotting the past data and visually
examining the appeared pattern:
 Trend - long-term movement in data
 Seasonality - short-term regular variations in
data
 Cycle – wavelike variations of more than one
year’s duration
 Irregular variations - caused by unusual
circumstances
 Random variations - caused by chance

3-19
 Forecast Variations
Figure 3.1

Irregular
variation

Trend

Cycles

90
89
88
Seasonal variations

3-20
Naive Forecasts
Uh, give me a minute....
We sold 250 wheels last
week.... Now, next week
we should sell....

The forecast for any period equals

the previous period’s actual value.

3-21
 Naïve Forecasts
 Simple to use
 Virtually no cost
 Quick and easy to prepare
 Data analysis is nonexistent
 Easily understandable
 Cannot provide high accuracy
 Can be a standard for accuracy

3-22
 Uses for Naïve Forecasts
 Stable time series data
 F(t) = A(t-1)
 Seasonal variations
 F(t) = A(t-n)
 Data with trends
 F(t) = A(t-1) + (A(t-1) – A(t-2))

3-23
 Averaging
 White noise and unimportant factors
 Minor variation and large variation
 Random variation and Real variation
 Averaging to smooth data fluctuation
 Three types of data changes

Ideal change Step change Gradual change

3-24
 Techniques for Averaging

 Moving average
 Weighted moving average
 Exponential smoothing

3-25
 Moving Averages
A technique that averages a number of recent
actual values, updated as new values become
available.
At-1 + At-2….. + At-n
Ft = MAn=
n
Ft = forecast for time period t
MAn = n period moving average
At-1 = actual value in period t-1
n = number of periods ( data points )

3-26
 Moving Averages
Example :
Compute a 3-period moving average forecast
given demand for shopping carts for the last
five periods as shown:
5 4 3 2 1 period
6
??? 41 40 43 40 42 demand

At-1 +At-2 ….+ At-n

Ft = MAn=
n
A5 + A4 + A3
t=6 F6 = MA3 = = 41.33
3

3-27
 Moving Averages
Example ( cont.) forecast

5 4 3 2 1 period
7 6
??? 41.33 41 40 43 40 42 demand
t=7 actual
6 5 4 3 2 1 period
7
??? 38 41 40 43 40 42 demand
A6 + A 5 + A 4 = 39.67
F7 = MA3=
3
forecast

6 5 4 3 2 1 period
7
39.67 38 41 40 43 40 42 demand
3-28
 Moving Averages

Error Forecast Actual Period

  #N/A 42 1

  #N/A 40 2

  #N/A 43 3

-1.6666667 41.666667 40 4

0 41 41 5

-2.3333333 41.333333 39 6

  40   7

3-29
Demand Moving Averages

Data

MA3

MA5

Period

3-30
 Moving Average
Actual
MA5
47
45
43
41
39
37 MA3
35
1 2 3 4 5 6 7 8 9 10 11 12

3-31
 Moving Averages

Decreasing the number of data points in a moving average

technique, increase the weight of more recent values

As data points in a moving average technique increased,

the sensitivity ( responsiveness ) of the average to new
values decreased.

If responsiveness is required, average with few data

points should be used,
 Moving Averages

It is easy to compute All values in the average are

It is easy to understand weighted equally, the oldest
value has the same weight
as the most recent value

But
 Weighted Moving Averages

 More recent values in a series are given

more weight in computing the forecast.
 Trial and error used to find the suitable
weighting scheme
 The sum of all weights must be = 1
 Weighted average technique is more
reflective of the most recent occurrences.

3-34
 Weighted Moving Averages
Example :
Compute a 4-period weighted moving average forecast given demand
for shopping carts for the last periods values and weights as shown:

5 4 3 2 1 period
6
??? 41 40 43 40 42 demand
F6 = W5*A5 + W4*A4 + W3*A3 + W2*A2
0.4 0.3 0.2 0.1 0
F6 = 0.4(41) + 0.3(40) + 0.2(43) + 0.1(40) = 41 weight
If the actual value of F6 is 39 then
F7 = 0.4(39) + 0.3(41) + 0.2(40) + 0.1(43) = 40.2

3-35
 Exponential Smoothing

•The most recent observations

might have the highest
predictive value.
•Therefore, we should give
more weight to the more
recent time periods when
forecasting.

3-36
 Exponential Smoothing

Weighted averaging method based on

previous forecast plus a percentage of
the forecast error

Ft = Ft-1 + (At-1 - Ft-1)

3-37
Example 3 - Exponential Smoothing

Period Actual Alpha = 0.1 Error Alpha = 0.4 Error

1 42
2 40 42 -2.00 42 -2
3 43 41.8 1.20 41.2 1.8
4 40 41.92 -1.92 41.92 -1.92
5 41 41.73 -0.73 41.15 -0.15
6 39 41.66 -2.66 41.09 -2.09
7 46 41.39 4.61 40.25 5.75
8 44 41.85 2.15 42.55 1.45
9 45 42.07 2.93 43.13 1.87
10 38 42.36 -4.36 43.88 -5.88
11 40 41.92 -1.92 41.53 -1.53
12 41.73 40.92

3-38
 Picking a Smoothing Constant
Actual
50
.4
D e m a nd

.1
45

40

35
1 2 3 4 5 6 7 8 9 10 11 12
Period

Selecting a smoothing constant α is a

matter of judgment or trial and error
3-39
 Techniques for Trend
Nonlinear Trends

Parabolic

Exponential

Growth

3-40
 Techniques for Trend
Linear Trends
8
7 7
6 6
5 5
4 4
3 3
2 2
1 1
0 0

3-41
 Linear Trend Equation
Ft

Ft = a + bt

0 1 2 3 4 5 t
 Ft = Forecast for period t
 t = Specified number of time periods
 a = Value of Ft at t = 0
 b = Slope of the line

3-42
Calculating a and b

n  (ty) -  t  y
b =
n t 2 - (  t) 2

 y - b t
a =
n

3-43
Linear Trend Equation Example

t y
2
W eek t S a le s ty
1 1 150 150
2 4 157 314
3 9 162 486
4 16 166 664
5 25 177 885

2
 t = 15  t = 55  y = 812  ty = 2 4 9 9
2
(  t) = 2 2 5

3-44
 Linear Trend Calculation
5 (2499) - 15(812) 12495-12180
b = = = 6.3
5(55) - 225 275 -225

812 - 6.3(15)
a = = 143.5
5

y = 143.5 + 6.3t

3-45
 Techniques for Seasonality
 Seasonal variations
 Regularly repeating movements in series values that can be tied to
recurring events.
 It is applied to monthly, weekly, daily and other regularly recurring
patterns in data.
 Seasonality is
 The amount that the actual values deviate from the average value of
a series.
 Seasonal relative
 Percentage of average or trend
 Centered moving average
 A moving average positioned at the center of the data that were
used to compute it.

3-46
 MODELS FOR SEASONALITY

THERE ARE TWO DIFFERENT MODELS OF SEASONALITY:

ADDITIVE MODEL:
SEASONALITY IS EXPRESSED AS QUANTITY (E.G., 20
UNITS), WHICH IS ADDED TO OR SUBTRACTED FROM
THE SERIES AVERAGE (OR TREND).

MULTIPLICATIVE MODEL:
SEASONALITY IS EXPRESSED AS PERCENTAGE TO BE
MULTIPLIED BY THE AVERAGE OF THE SERIES (OR
TREND), IT IS ALWAYS CALLED SEASONAL RELATIVES
OR SEASONAL INDEX.

3-47
SEASONALITY AND AVERAGE MODELS

Average

Demand = average + seasonality

Average

Demand = average * seasonal relative

3-48
 SEASONALITY AND TREND MODELS

Trend

Demand = trend + seasonality

Demand = trend * seasonal relative

Trend

3-49
 Seasonal relatives
Seasonal relatives are used in two different
ways in forecasting to:

• Incorporate seasonality in a forecast

• Deseasonalize the data

3-50
Incorporating seasonality into forecast

Incorporating seasonality into forecast is

useful when demand has both trend (or
average) and seasonal components,
which is conducted as following:
1. Obtain trend estimates for the desired
periods using trend equation.
2. Add seasonality to the trend by
multiplying (assuming multiplicative
model is appropriate) these trend
estimates by the corresponding seasonal
relative.
3-51
Incorporating seasonality into forecast
 Example
A furniture manufacturer wants to predict quarterly demand for
certain seats for periods 15 and 16, which happen to be the
second and the third quarters for a particular year. The series
consists of both trend and seasonality. The trend portion of the
demand is projected using the trend equation Ft = 124 + 7.5 t.
Quarter relatives are Q1 = 1.2 , Q2 = 1.1, Q3 = 0.75 and Q4 = 0.95.
use this information to predict demand for periods 15 and 16.
 Solution
The trend values at t = 15 and t = 16 are:
F15 = 124 + 7.5 (15) = 236.5
F16 = 124 + 7.5 (16) = 244.0
Multiplying the trend value by the appropriate quarter relatives yield
a forecast that includes both trend and seasonality. Given that t
= 15 is a second quarter and t = 16 is a third quarter, the forecast
will be:
Period 15: 236.5 (1.1) = 260.15 Period 16: 244.0 (0.75) = 183.00
3-52
 Example (cont)
Use the following information to deseasonalize
sales for quarters 1 through 8
Sales Quarter Period Des. Sal.
132 1 1 110
140 2 2 127.27
.Q. R
146 3 3 194.67
1.20
153 4 4 161.05
1.10
160 1 5 133.33
0.75
168 2 6 152.73
0.95
176 3 7 234.67
185 4 8 194.74
3-53
 Calculating seasonal relatives

 Example
The manager of a parking lot has
computed daily relatives for the number of
cars per day in the lot. The computations
are repeated here (about three weeks are
shown for illustration). A seven period
centered moving average is used because
there are seven days (seasons) per week

3-54
Calculation of seasonal relatives (cont.)

Cars/MA7 Centered MA7 Cars Day

67 Tues

75 Wed

82 Thur

1.36 71.86 98 Fri

1.27 70.86 90 Sat

0.51 70.57 36 Sun

0.77 71 55 Mon

0.84 71.14 60 Tues

1.03 70.57 73 Wed

1.19 71..14 85 Thur

1.4 70.71 99 Fri

1.21 71.29 86 Sat

0.56 71.71 40 Sun

0.72 72 52 Mon

0.89 71.57 64 Tues

1.06 71.86 76 Wed

1.2 72.43 87 Thur

1.33 72.14 96 Fri

88 Sat

44 Sun

50 Mon

3-55
 Seasonal Relatives (cont.)
The estimated relatives will be:
 Monday: (0.77 + 0.72)/2 = .745
 Tuesday: (0.84 + 0.89)/2 = .865
 Wednesday: (1.03 + 1.06)/2 = 1.045
 Thursday: (1.19 + 1.2)/2 = 1.195
 Friday: (1.36 + 1.4 + 1.33)/3 = 1.36
 Saturday: (1.27 + 1.21)/2 = 1.24
 Sunday: (0.51 + 0.56)/2 = 0.535
Note
The sum of the relatives must equal the number of periods
(i.e., 7 in this example). If it is not, you have to multiply by
a correction factor. In this example the sum is 6.985,
therefore you have to multiply each factor by (7/6.985)

3-56
 Seasonal Relatives
The number of periods needed in a centered moving
average is equal to the number of “seasons” involved
 With monthly data, a 12-period moving average is
needed.
 With daily data, a 7-period moving average is needed.
 With quarterly data, a 4-period moving average is needed
Note:
When the number of period is even, one additional step is
needed, because the middle of an even period falls
between two periods. The additional step requires taking
two-period moving average of the even-numbered
centered moving average. (See the following example)

3-57
 Seasonal relatives

 Example (even period)

Obtain estimates of quarter relatives for these data:
4 3 2 1 Year

1 4 3 2 1 4 3 2 1 4 3 2 1 Quarter

58 88 84 54 45 71 60 36 28 46 35 18 14 Demand

Note:
The number of seasons is 4 (an even number); therefore, a
4-centered moving average will be taken and then a 2-
centered moving average is calculated to adjust the data.
3-58
 Seasonal Relatives
D/MA2 MA2 MA4 Demand (D) Quarter Year
14 1 1
18
28.25
2
1.17 30 35
31.75 3
1.35 34 46 4
36.25
0.71 39.38 28 1 2
42.50
0.79 45.63 36
48.75
2
1.18 50.88 60
53.00 3
1.29 55.25 71
57.50 4
0.74 60.50 45
63.50 1 3
0.82 65.63 54
67.75
2
1.21 69.38 84
71.00 3
88 4
58 1 4
 Seasonal relatives

Average relatives
Quarter1 = (0.71+0.74)/2 = 0.725
Quarter2 = (0.79+ 0.82)/2 = 0.805
Quarter3 = (1.17 + 1.18 + 1.21)/3 = 1.187
Quarter4 = (1.35 + 1.29)/2 = 1.32
The sum of these relatives is 4.037. Multiplying
each by 4/4.037 will standardize relatives making
their total equal 4. The resulting relatives are:
Quarter1 = 0.718, Quarter2 = 0.798
Quarter3 = 1.176, Quarter4 = 1.308 3-60
 Deseasonalize the data

 To deseasonalize the data, divide the

demand of each period by the
corresponding seasonal relatives.
 For a good forecast, use the
deseasonalized data to fit the trend
equation instead of the original data. Then
multiply the trend by the associated relative
of the period.

3-61
 Deseasonalize the data

Actual
120 – data (y1)
Sales (millions of dollars)

100 –

80 – Deseasonalized
data (dt)
60 –

40 –

20 –
| | | | | | | | | | | | | | | |
t
1 2 3 4 1 2 3 4 1 2 3 4 1 2 3 4
1998 1999 2000 2001

3-62
 Associative Forecasting
Associative models assume that there is a
causational relationship between the
variable of interest and other variables
called predictors
Examples:
• Price of beef and price of chicken
• Crop yields and soil condition
• Crop yields and timing of water application
• Profits and sales
• Price of products and energy cost
 Associative Forecasting

Def.
 Predictor variables - used to predict
values of variable of interest
 Regression - technique for fitting a line
to a set of points
 Least squares line - minimizes sum of
squared deviations around the line

3-64
 Simple Linear Regression
 The simplest and most widely used form of regression
involves a linear relationship between two variables.
 The objective in linear regression is to obtain an equation
of a straight line that minimizes the sum of squared
vertical deviations of data points from the line (i.e., the
least squares criterion)
 This least squares line has the form:
Y = a + bX
Where:
Y = predicted (dependent or response) variable
X = predictor (independent or explanatory) variable
a = intercept
b = slope (change rate)

3-65
Simple Linear Regression

n  (xy) -  x  y
b =
n x2 - (  x)2

 y - b x
a =
n

n = number of paired observations

3-66
 Simple Linear Regression
Sales Profits
7 15
2 10
6 13
4 15
14 25
15 27
16 24
12 20
14 27
20 44
15 34
7 17
3-67
 Simple Linear Regression

x2 xy Y X n  (xy) -  x  y
b =
n x2 - (  x)2
49 105 15 7
4 20 10 2
36 78 13 6
16 60 15 4 1.72870662 =b
196 350 25 14
225 405 27 15
 y - b x
256 384 24 16 a =
144 240 20 12 n
196 378 27 14 3.5675563 =a
400 880 44 20
225 510 34 15
49 119 17 7 Y = 3.567 + 1.728X
1796 3529 271 132
 Simple Linear Regression

Profit = 3.567 + 1.728(sales)

If you want to
predict profit at

PROFITS
sales point = 10.3

Profit = 21.365

3-69
 PREDICTORES

Three conditions are required for predictor to

be valid:
1. The relationship between movements of an
indicator and movements of the variable
should have a logical explanation.
2. Movements of the indicator must precede
movements of the dependent variable by
enough time so that the forecast isn’t outdated
before it can be acted upon.
3. A fairly high correlation should exist between
the two variables.
 CORRELATION

A measure of the strength and direction of relationship

between two variables

n  (xy) -  x  y
r =
2 2 2 2
(( ) ( ))
√ n  x - (  x)
-1< r < +1
n y - (  y)

•Correlation r close to zero indicate weak linear relationship between two variables

•Correlation r close to +1 indicate strength linear relationship between two variables

(directly change)
•Correlation r close to -1 indicate strength linear relationship between two variables
(inversely change)
3-71
 Simple Regression Analysis
 Assumptions
 Variations around the line are random
 Deviations around the line normally distributed
 Predictions are being made only within the range
of observed values
 For best results:
 Always plot the data to verify linearity
 Check for data being time-dependent
 Small correlation may imply that other variables
are important
3-72
 Simple Regression Analysis

 Weakness of regression
 Simple linear regression applies only to linear
relationship with one independent variable .
 One needs a considerable amount of data to
establish the relationship – in practice, 20 or
more observations.
 All observations are weighted equally.

3-73
 Simple Regression Analysis
EXAMPLE: Unemp. units
Sales of new houses and period
rate sold
three-month lagged
unemployment are 7.2 20 1

shown in the following 4 41 2

table. 7.3 17 3
Determine if 5.5 35 4
unemployment levels 6.8 25 5
can be used to predict 6 31 6
demand for new houses. 5.4 38 7
If so, derive a predictive 3.6 50 8
equation. 8.4 15 9
7 19 10
9 14 11
3-74
Simple Regression Analysis

3-75
Simple Regression Analysis

Unemp. rate
Simple Regression Analysis

3-77
 Simple Regression Analysis
Y² X² XY x y t
400 51.84 144 7.2 20 1
1681 16 164 4 41 2
289 53.29 124.1 7.3 17 3
1225 30.25 192.5 5.5 35 4
625 46.24 170 6.8 25 5
961 36 186 6 31 6
1444 29.16 205.2 5.4 38 7
2500 12.96 180 3.6 50 8
225 70.56 126 8.4 15 9
361 49 133 7 19 10
196 81 126 9 14 11

9907 476.3 1750.8 70.2 305

ΣY² ΣX² ΣXY ΣX ΣY
r = ((11*1750.8)-(305*70.2))/(SQRT(((11*476.3)-(70.2*70.2))*((11*9907)-
(305*305))))
r = - 0.97 High Negative Relationship 3-78
 Simple Regression Analysis

b =((11*1750.8)-(305*70.2))/(((11*476.3)-(70.2*70.2)))
b = - 6.91

a = 305 – ( - 6.91 ) / 11
a = 71.85

The regression equation will be

(Unit sold) = 71.85 - 6.91*(unemp. rate)

3-79
 Forecast Accuracy
 Error - difference between actual value and
predicted value
 Mean Absolute Deviation (MAD)
 Average absolute error
 Mean Squared Error (MSE)
 Average of squared error
 Mean Absolute Percent Error (MAPE)
 Average absolute percent error

3-80
 MAD, MSE, and MAPE

 Actual  forecast
MAD =
n
2
 ( Actual  forecast)
MSE =
n -1

 Actual  forecast / Actual*100)

MAPE =
n

3-81
 MAD, MSE and MAPE
 MAD
 Easy to compute
 Weights errors linearly
 MSE
 Squares error
 More weight to large errors
 MAPE
 Puts errors in perspective

3-82
 Example 10

Period Actual Forecast (A-F) |A-F| (A-F)^2 (|A-F|/Actual)*100

1 217 215 2 2 4 0.92
2 213 216 -3 3 9 1.41
3 216 215 1 1 1 0.46
4 210 214 -4 4 16 1.90
5 213 211 2 2 4 0.94
6 219 214 5 5 25 2.28
7 216 217 -1 1 1 0.46
8 212 216 -4 4 16 1.89
-2 22 76 10.26

MAD= 2.75
MSE= 10.86
MAPE= 1.28

3-83
 Sources of Forecast errors
 Model may be inadequate
 Irregular variations
 Incorrect use of forecasting technique

3-84
 Controlling the Forecast
 Control chart
 A visual tool for monitoring forecast errors
 Used to detect non-randomness in errors
 Forecasting errors are in control if
 All errors are within the control limits
 No patterns, such as trends or cycles, are
present

3-85
 Control chart
 The upper control limit

UCL = 0 + z . MSE

 Lower Control Limit

LCL = 0 – z. MSE

Where: z is the number of standard deviations. For 95%

confidence use z = 2, for 99% confidence use z = 3
 Control Chart

Abnormal error Out of

error control
UCL

Mean = 0
Normal error
due to chance
LCL
Abnormal error

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
period
 Tracking Signal
•Tracking signal
The ratio of cumulative error to MAD

Tracking signalt =
 Actualt(- )forecastt
MADt
•The tracking signal is computed period by period and it
can be positive or negative, a value of zero would be
ideal; limits of ± 4 or ± 5 are often used for a range of
acceptable values of tracking signal. If a value is out of
the range, then there is a bias in the forecast.
•Bias – Persistent tendency for forecasts to be
Greater or less than actual values.
 Tracking signal
 The following table shows the calculation of the tracking signal (TS).
Note that all values are within ± 4
TS MADt Cumulative Cumulative e e =A-F Forecast Sales Month
e Error (Error) (F) (A)

-1 4 4 -4 4 -4 47 43 1

0.55 5.5 11 3 7 7 44 51 2

1.4 5 15 7 4 4 50 54 3

2 4 16 8 1 1 50 51 4

0.71 4.2 21 3 5 -5 54 49 5

0.26 3.8 23 1 2 -2 48 46 6
 Choosing a Forecasting
Technique
 No single technique works in every
situation
 Two most important factors
 Cost
 Accuracy
 Other factors include the availability of:
 Historical data
 Computers
 Time needed to gather and analyze the data
 Forecast horizon
3-90
 Operations Strategy
 Forecasts are the basis for many decisions
 Work to improve short-term forecasts
 Accurate short-term forecasts improve
 Profits
 Lower inventory levels
 Reduce inventory shortages
 Improve customer service levels
 Enhance forecasting credibility

3-91
 Supply Chain Forecasts
 Sharing forecasts with supply chain can
 Improve forecast quality in the supply chain
 Lower costs
 Shorter lead times

3-92
Exponential Smoothing

3-93
Linear Trend Equation

3-94
Simple Linear Regression

3-95