Forecasting

T. A. Pai Management Institute

Manipal

Forecasting
The Forecast of future demand forms the basis for
all strategic and planning decision in a supply
chain.
The following are the few functional areas that are
based on demand forecast
Production: Scheduling, inventory control, aggregate
planning

Marketing: Sales-force allocation, promotions, new

product introduction

Finance: Plant/equipment investment, budgetary

planning

Personnel: Workforce planning, hiring, layoffs

Some Applications of Forecasting

Sales Forecasting
Forecasting the Need for Spare Parts
Forecasting Production Yields
Forecasting Economic Trends
Forecasting Staffing Needs

Characteristics of Forecast
Forecasts almost always contain errors and should
include both the expected value and a measure of
forecast error
Long term forecasts are usually less accurate than
short term forecast
Aggregate forecasts are usually more accurate than
disaggregate forecasts
Common sense compatibility of forecasting

Long term forecasts are usually

less accurate than short term
forecast forecasts
Sales
Forecast
Sales
Past
Sales
Actual
Sales

Year

Aggregate forecasts are usually

more accurate than disaggregate
forecasts
Item
A
B
C
D
E

Total

Absolute
Forecast Actual
Demand Demand Error |A-F|
200
80
150
300
150

215
79
128
290
162

880

Aggregate Error:

874
6

15
1
22
10
12
60

Components of a Forecast
The following factors may be related to the
demand forecast besides numerous other
factors
Past demand
Planned Promotional efforts
State of the economy
Planned price discounts
Actions competitors have taken

Components of a Forecast
Observed demand

Systematic component +

Random
component

Components of a Forecast
Systematic component of Demand :
Measures the expected value of demand and
consists of

Level/ Horizontal : The current demand

Trend : The rate of growth or decline in
demand for the next period

Seasonality : The predictable seasonal

fluctuations in demand

Demand Patterns
A time series is the repeated
observations of demand for a service or
product in their order of occurrence
There are five basic time series
patterns
Horizontal
Trend

Seasonal
Cyclical
Random

Demand Patterns

Quantity

Figure 13.1 Patterns of Demand

Time
(a) Horizontal: Data cluster about a horizontal line

Demand Patterns

Quantity

Figure 13.1 Patterns of Demand

Time
(b) Trend: Data consistently increase or decrease

Demand Patterns
Figure 13.1 Patterns of Demand

Quantity

Year 1

Year 2
|
J

|
F

|
M

|
A

|
M

|
|
J
J
Months

|
A

|
S

|
O

|
N

|
D

(c) Seasonal: Data consistently show peaks and valleys

Demand Patterns

Quantity

Figure 13.1 Patterns of Demand

|
1

|
2

|
3

|
4

|
5

Years
(d) Cyclical: Data reveal gradual increases and
decreases over extended periods

|
6

Demand Management Options

Demand Management
The

process of changing demand patterns

using one or more demand options

Demand Management Options

Complementary Products
Promotional Pricing
Prescheduled Appointments
Reservations
Revenue Management
Backlogs

Backorders and Stockouts

Key Decisions
Deciding what to forecast
Level

of aggregation

Cluster several similar services or goods

Have similar demand requirements and common
processing, labor and material requirement
Derive forecasts for individual items

Units

of measurement

Choosing a forecasting system

Choosing the type of forecasting technique
Judgment

and qualitative methods

Causal methods
Time-series analysis

Use of Forecasting: Operations

Decisions
Time
Horizon

Accuracy
Required

Number of
Forecasts

Long

Medium

Single or few Top

Qualitative
or causal

Long

Medium

Single or few Top

Qualitative
and causal

Medium

High

Few

Middle

Causal and
time series

Short

Highest

Many

Lower

Time series

Inventory
management Short

Highest

Many

Lower

Time series

Process
design
Capacity
planning,
facilities
Aggregate
planning
Scheduling

Management Forecasting
Level
Method

11-18

13 18

Use of Forecasting: Marketing, Finance &

HR
Time
Horizon

Accuracy
Required

Number of
Forecasts

Long

Medium

Single or few Top

Qualitative

Short

High

Many

Middle

Time series

New product
introduction Medium

Medium

Single

Top

Qualitative
and causal

Cost
estimating

Short

High

Many

Lower

Time series

Capital
budgeting

Medium

Highest

Few

Top

Causal and
time series

Long-range
marketing
programs
Pricing
decisions

Management Forecasting
Level
Method

11-19

13 19

Judgment Methods
Other methods (casual and time-series)
require an adequate history file, which
might not be available
Judgmental forecasts use contextual
knowledge gained through experience
Sales force estimates
Executive opinion is a method in which
opinions, experience, and technical
knowledge of one or more managers are
summarized to arrive at a single forecast
Delphi method Used when neither
historical data nor relevant expertise is
available within the organization.
Consensus of Experts

Judgment Methods
Market research is a systematic approach
used to determine external customer
interest through data-gathering surveys
Delphi method is a process of gaining
consensus from a group of experts while
maintaining their anonymity

Useful when no historical data are available

Can be used to develop long-range forecasts
and technological forecasting

FORECASTING FOR OPERATIONS..

..QUALITATIVE/JUDGEMENTAL
METHODS
Used when we dont have past data,historical records or
*
when statistical methods
have no validity.
Used for long-range predictions / new product
developments /market strategies and facilities planning in
connection with the above.

22

7/22/02
1/5/2017

FORECASTING FOR OPERATIONS..

..QUALITATIVE/JUDGEMENTAL
A.DELPHI METHOD:
*

Used as a tool for technological forecasting.

The objective is to probe the future in hopes of anticipating

new processes and products in the rapidly changing
environment of todays culture and economy.

23

POTENTIAL DOMINANCE OF THE MOST

PRESTIGIOUS/VERBAL members of the panel are
eliminated.
7/22/02

1/5/2017

FORECASTING FOR OPERATIONS..

..DELPHI, General Procedure contd
3On the basis of the above, the coordinator writes out a
new set of questions* to the experts that may include
feedback from other experts.
4.This is repeated for a few rounds, till the coordinator is
satisfied with the overall prediction synthesised by the
experts.
The coordinator has to be talented for the job.
Technique was first developed by RAND Corporation.

24

7/22/02
1/5/2017

1.FORECASTING FOR OPERATIONS..

..DELPHI, Shortcomings
Takes very long, sometimes a year, for the repetition of
*
rounds.
Experts may change and confusion can result due to
long gaps.

25

7/22/02
1/5/2017

Time Series Methods

1. Naive Forecast or Last Value
In a Naive Forecast the forecast for the
next period equals the demand for the
current period
Forecast = Last Value
Forecast Ft+1 = Dt
The Last- Value forecasting method ignores all the data points
in a time series except the last one. It uses last value as the
basis for the forecast of what the next data point will be

When conditions are changing rapidly, it may be that last value

is the only relevant data point for forecasting the next value
under current conditions.

Forecasting Error
For any forecasting method, it is important to
measure the accuracy of its forecasts.
Forecast error is simply the difference found
by subtracting the forecast from actual
demand for a given period, or
Et = Dt Ft
where
Et = forecast error for period t
Dt = actual demand in period t
Ft = forecast for period t

Calculating Forecast Errors

Et = Dt Ft

CFE = Et
CFE
E= n

Et = forecast error for period t

Dt = actual demand in period t
Ft = forecast for period t
Cumulative sum of forecasting error
Mean bias

|Et |
MAD =
n

MSE =

where

Et2
n

Mean absolute deviation

Mean square error

Time Series Methods

2. Simple Moving Averages
Estimating the average: Simple Moving
Averages
Used

to estimate the average of a demand

time series and thereby remove the effects
of random fluctuation
Most useful when demand has no
pronounced trend or seasonal influences
The stability of the demand series generally
determines how many periods to include

2. Simple Moving Averages

Specifically, the forecast for period t + 1 can
be calculated at the end of period t (after the
actual demand for period t is known) as

Forecast = Average of last n values

Dt + Dt-1 + Dt-2 + + Dt-n+1
Sum of last n demands
Ft+1 =
=
n
n
where
Dt = Actual demand in period t
n = total number of periods in the average
Ft+1 = Forecast for period t +1

3.Weighted Moving Averages

Each historical demand in the average can have its
own weight, provided that the sum of the weights
equals 1.0.
The average is obtained by multiplying the weight of
each period by the actual demand for that period,
and then adding the products together:
Ft+1 = W1D1 + W2D2 + + WnDt-n+1

A three-period weighted moving average model with

the most recent period weight of 0.50, the second
most recent weight of 0.30, and the third most
recent might be weight of 0.20
Ft+1 = 0.50Dt + 0.30Dt1 + 0.20Dt2

4.Exponential Smoothing
A sophisticated weighted moving average that
calculates the average of a time series by
giving recent demands more weight than
earlier demands
Requires only three items of data

The last periods forecast

The demand for this period

A smoothing parameter, alpha (), where 0 1.0

The equation for the forecast is

Forecast = (Last value) +(1- ) (Last Forecast)
Ft+1 = (Demand last period) + (1 )(Forecast calculated last period)

= Dt + (1 )Ft

4.Exponential Smoothing
The emphasis given to the most recent
demand levels can be adjusted by changing
the smoothing parameter
Larger values emphasize recent levels of
demand and result in forecasts more
responsive to changes in the underlying
average

A larger value (say = 0.5) is needed if significant

changes in the conditions are occurring frequently

Smaller values treat past demand more

uniformly and result in more stable forecasts

A small value (say = 0.1) is appropriate if condition

are remaining relatively stable

Exponential smoothing is simple and requires

minimal data

5.Using Trend-Adjusted
Exponential Smoothing

Including a Trend
A trend in a time series is a systematic
increase or decrease in the average of
the series over time

The forecast can be improved by

calculating an estimate of the trend
Trend-adjusted exponential smoothing
requires two smoothing constants

5.Using Trend-Adjusted
Exponential Smoothing
For each period, we calculate the average and
the trend:
At = (Demand this period)
+ (1 )(Average + Trend estimate last period)
= Dt + (1 )(At1 + Tt1)

Tt = (Average this period Average last period)

+ (1 )(Trend estimate last period)
= (At At1) + (1 )Tt1
Ft+1 = At + Tt
where
At =
Tt =
=
=
Ft+1 =

exponentially smoothed average of the series in period t

exponentially smoothed average of the trend in period t
smoothing parameter for the average, with a value between 0 and 1
smoothing parameter for the trend, with a value between 0 and 1
forecast for period t + 1

CLASSIFICATION OF EXPONENTIAL SMOOTHING MODELS

SAMPLE CALCULATION, ( =0.2, =0.1)
MONTH

ACTUAL
DEMAND

Initial JAN
FEB
MAR
APR
MAY
JUN
JUL

, Dt

0.00
19.36
25.45
19.73
21.48
20.77
25.42
-

Average

At
20.00
19.87
20.98
20.81
20.99
21.02
21.96
-

TREND

Tt
0.000
-0.013
0.099
0.072
0.084
0.077
0.1643
-

FORECAST

t t+1

0.00
20.00
19.86
21.08
20.88
21.10
21.10
22.13

AMAR = (0.2 x 19.73 ) + 0.8 x (20.98 + 0.099) = 20.807

TMAR = 0.1 x (20.807 -20.977 ) + (0.9 x 0.099) = 0.0721
Forecast,April = 2.8071 + 0.0721 = 20.8792

6.Seasonal Patterns
Seasonal patterns are regularly
repeated upward or downward
movements in demand measured in
periods of less than one year (hours,
days, weeks, months or quarter)
Account for seasonal effects by using
one of the techniques already described
but to limit the data in the time series
to those periods in the same season
This approach accounts for seasonal
effects but discards considerable
information on past demand

Multiplicative Seasonal Method

Multiplicative seasonal method, whereby seasonal
factors are multiplied by an estimate of the average
demand to arrive at a seasonal forecast

1. For each year, calculate the average demand

for each season by dividing annual demand
by the number of seasons per year
2. For each year, divide the actual demand for
each season by the average demand per
season, resulting in a seasonal index for each
season
3. Calculate the average seasonal index for each
season using the results from Step 2
4. Calculate each seasons forecast for next year

Example 1 for Multiplicative

Seasonal Method
EXAMPLE 13.5
The manager of the Stanley Steemer carpet cleaning company
needs a quarterly forecast of the number of customers
expected next year. The carpet cleaning business is seasonal,
with a peak in the third quarter and a trough in the first
quarter. Following are the quarterly demand data from the
past 4 years:
Quarter

Year 1

Year 2

Year 3

Year 4

45

70

100

100

335

370

585

725

520

590

830

1160

100

170

285

215

Total

1000

1200

1800

2200

The manager wants to forecast customer demand for each quarter

of year 5, based on an estimate of total year 5 demand of 2,600
customers

Example 1 for Multiplicative

Seasonal Method
Quarter
1
2
3
4
Total
Average

Year 1
45
335
520
100
1000
250

Year 2
70
370
590
170
1200
300

Year 3
100
585
830
285
1800
450

Year 4
100
725
1160
215
2200
550

Year 5

Quarter
1
2
3
4

Year 1
0.18
1.34
2.08
0.40

Year 2
0.23
1.23
1.97
0.57

Year 3
0.22
1.30
1.84
0.63

Year 4
0.18
1.32
2.11
0.39

Average Index
0.20
1.30
2.00
0.50

2600
650

Example 2 for Multiplicative

Seasonal Method
Suppose the multiplicative seasonal method is being used
to forecast customer demand. The actual demand and
seasonal indices are shown below.
Year 1

Year 2

Average
Index

Quarter

Demand

Index

Demand

Index

100

0.40

192

0.64

0.52

400

1.60

408

1.36

1.48

300

1.20

384

1.28

1.24

200

0.80

216

0.72

0.76

Average

250

300

Choosing a Time-Series Method

Forecast performance is determined by
forecast errors
Forecast errors detect when something is
going wrong with the forecasting system

Forecast errors can be classified as either

bias errors or random errors
Bias errors are the result of consistent
underlying deviations in the data
Random error results from unpredictable
factors that cause the forecast to deviate
from the actual demand

Forecasting Error
For any forecasting method, it is important to
measure the accuracy of its forecasts.
Forecast error is simply the difference found
by subtracting the forecast from actual
demand for a given period, or
Et = Dt Ft
where
Et = forecast error for period t
Dt = actual demand in period t
Ft = forecast for period t

Calculating Forecast Errors

Et = Dt Ft

CFE = Et
CFE
E= n

Et = forecast error for period t

Dt = actual demand in period t
Ft = forecast for period t
Cumulative sum of forecasting error
Mean bias

|Et |
MAD =
n

MSE =

where

Et2
n

Mean absolute deviation

Mean square error

Measures of Forecast Error

MAPE =

(Et E )2

Standard Deviation of error (S)

n1

({|Et |/Dt})(100)
n

Mean absolute percent error

Calculating Forecast Errors

EXAMPLE 13.6

The following table shows the actual sales of upholstered

chairs for a furniture manufacturer and the forecasts
made for each of the last eight months. Calculate CFE,
MSE, , MAD, and MAPE for this product.
Month
t
1
2
3
4
5
6
7
8

Demand Forecast Error Error2

Dt
200
240
300
270
230
260
210
275

Ft
225
220
285
290
250
240
250
240
Total

Et

Et2

Absolute Absolute %
Error
Error
|Et|
(|Et|/Dt)(100)

Calculating Forecast Errors

The following table shows the actual sales of upholstered
chairs for a furniture manufacturer and the forecasts made
for each of the last eight months. Calculate CFE, MSE, ,
MAD, and MAPE for this product.

Month

Demand

Forecast

Error

t
1
2
3
4
5
6
7
8

Dt
200
240
300
270
230
260
210
275

Ft
225
220
285
290
250
240
250
240
Total

Et
-25
20
15
-20
-20
20
-40
35
-15

Error2

E t2
625
400
225
400
400
400
1600
1225
5275

Absolute Absolute %
Error
Error
|Et|
(|Et|/Dt)(100)
25
12.5
20
8.3
15
5.0
20
7.4
20
8.7
20
7.7
40
19.0
35
12.7
195
81.4

S
(Et E )2
534.7656
478.5156
284.7656
328.5156
328.5156
478.5156
1453.516
1359.766
5246.9

Calculating Forecast Errors

SOLUTION

Using the formulas for the measures, we get

Cumulative forecast error (bias):

CFE = Et

CFE = 15

Average forecast error (mean bias):

E=

CFE
n

-15 = 1.875
8

Mean squared error:

Et2
5,275 = 659.4
MSE = n =
8

Calculating Forecast Errors

(Et E )2
n1

[Et (1.875)]2
n1

= 27.4

Mean absolute deviation:

|Et |
195
=
= 24.4
MAD =
n
8
Mean absolute percent error:
(|Et |/ Dt)(100)
81.3%
=
= 10.2%
MAPE =
n
8

Calculating Forecast Errors

A CFE of 15 indicates that the forecast has a slight bias to
overestimate demand.
The MSE, , and MAD statistics provide measures of forecast
error variability.
A MAD of 24.4 means that the average forecast error was 24.4
units in absolute value.
The value of of 27.4, indicates that the sample distribution
of forecast errors has a standard deviation of 27.4 units.
A MAPE of 10.2 percent implies that, on average, the forecast
error was about 10 percent of actual demand.
These measures become more reliable as the number of
periods of data increases.

Tracking Signals
A measure that indicates whether a method of
forecasting is accurately predicting actual
changes in demand
Useful when forecast systems are
computerized because it alerts analysts when
forecast are getting far from desirable limits

CFEt
Tracking signalt =
MADt
Each period, the CFE and MAD are updated to
reflect current error, and the tracking signal is
compared to some predetermined limits

Tracking Signals
The MAD can be calculated two ways

As a simple average of all absolute errors

As a weighted average determined by the

exponential smoothing method

MADt = |Et| + (1 )MADt-1

If forecast errors are normally distributed
with a mean of 0, the relationship between
and MAD is simple
= ( /2)(MAD) 1.25(MAD)

MAD = 0.7978 0.8

Control Limit
Spread (number
of MAD)

Equivalent
number of

Percent area within

control limits

+/- 1

+/- 0.8

57.62

+/- 1.5

+/- 1.2

76.98

+/- 2

+/- 1.6

89.04

+/- 2.5

+/- 2.0

95.44

+/- 3.0

+/- 2.4

98.36

Tracking Signals
Out of control

+2.0

Control limit

Tracking signal

+1.5
+1.0
+0.5
0
0.5
1.0

Control limit

1.5
|

Figure 13.7 Tracking Signal

10
15
20
Observation number

25

7.Causal methods : Linear

Regression
How do wages of employees depend on years
of experience, years of education and gender?
How does the current price of a stock depend
on its past values, as well as the current and
future values of a market index?
How does a companys current sales level
depend on its current and past advertising
levels, the advertising levels of its
competitors, the companys own past sales
levels, and the general level of the market?

7.Causal methods : Linear

Regression
How does the unit cost of producing an item
depend on the total quantity of items that
have been produced?
How does the selling price of a house depend
on such factors as the apprised value of the
house, the square footage of the house, the
number of bedrooms in a house and perhaps
others?
Regression analysis studies such relationships

Regression analysis purposes:

A) for understanding relationships
B) for making predictions

7.Causal methods : Linear

Regression
A dependent variable is related to one or more
independent variables by a linear equation
The independent variables are assumed to
cause the results observed in the past
Simple linear regression model is a straight
line
Y = a + bX
where
Y
X
a
b

= dependent variable
= independent variable
= Y-intercept of the line
= slope of the line

Linear Regression

Dependent variable

Deviation,
or error

Regression
equation:
Y = a + bX

Estimate of
Y from
regression
equation

Actual
value
of Y

Value of X used
to estimate Y
X

Independent variable
Figure 13.2 Linear Regression Line Relative to Actual Data

Linear Regression
The sample correlation coefficient, r

Measures the direction and strength of the

relationship between the independent variable and
the dependent variable.

The value of r can range from 1.00 r 1.00

The sample coefficient of determination, r2

Measures the amount of variation in the dependent

variable about its mean that is explained by the
regression line

The values of r2 range from 0.00 r2 1.00

The standard error of the estimate, syx

Measures how closely the data on the dependent variable

clusters around the regression line

Using Linear Regression

EXAMPLE 13.1
The supply chain manager seeks a better way to forecast
the demand for door hinges and believes that the demand
is related to advertising expenditures. The following are
sales and advertising data for the past 5 months:
Month

Sales (thousands of units)

Advertising (thousands of $)

264

2.5

116

1.3

165

1.4

101

1.0

209

2.0

The company will spend $1,750 next month on advertising for

the product. Use linear regression to develop an equation and
a forecast for this product.

Using Linear Regression

SOLUTION
We used POM for Windows to determine the best values of
a, b, the correlation coefficient, the coefficient of
determination, and the standard error of the estimate

a=
b=
r=
r2 =
syx =

8.135
109.229X
0.980
0.960
15.603

The regression equation is

Y = 8.135 + 109.229X

Using Linear Regression

The regression line is shown in Figure 13.3. The r of 0.98
suggests an unusually strong positive relationship
between sales and advertising expenditures. The
coefficient of determination, r2, implies that 96 percent
of the variation in sales is explained by advertising
expenditures.
Brass Door Hinge

Sales (000 units)

250

200

150

100

X
Data

X
Forecasts

50

1.0

2.0
Advertising ($000)

Figure 13.3 Linear Regression Line for the Sales and Advertising Data

Criteria for Selecting Methods

Criteria to use in selecting forecast method
and parameter choices include
1. Minimizing bias
2. Minimizing MAPE, MAD and MSE
3. Meeting managerial expectations to predict
changes in the components of demand
4. Minimizing the forecast error from the last period

Statistical performance measures can be used

1. For projections of more stable demand patterns, use
lower and values or larger n values
2. For projections of more dynamic demand patterns try
higher and values or smaller n values

Using Multiple Techniques

Combination forecasts are forecasts
that are produced by averaging
independent forecasts based on
different methods or different data or
both
Focus forecasting selects the best
forecast from a group of forecasts
generated by individual techniques

Forecasting as a Process
A typical forecasting process
Step 1: Adjust history file
Step 2: Prepare initial forecasts
Step 3: Consensus meetings and collaboration
Step 4: Revise forecasts
Step 5: Review by operating committee
Step 6: Finalize and communicate

Forecasting is not a stand-alone activity,

but part of a larger process

Forecasting as a Process

Adjust
history
file
1

Prepare
initial
forecasts
2

Consensus
meetings and
collaboration
3

Finalize
and
communicate
6

Review by
Operating
Committee
5

Revise
forecasts
4

Forecasting Principles
TABLE 13.2

SOME PRINCIPLES FOR THE FORECASTING PROCESS

Better processes yield better forecasts

Demand forecasting is being done in virtually every company, either
formally or informally. The challenge is to do it wellbetter than the
competition
Better forecasts result in better customer service and lower costs,
as well as better relationships with suppliers and customers
The forecast can and must make sense based on the big picture,
economic outlook, market share, and so on
The best way to improve forecast accuracy is to focus on reducing
forecast error
Bias is the worst kind of forecast error; strive for zero bias
Whenever possible, forecast at more aggregate levels. Forecast in
detail only where necessary
Far more can be gained by people collaborating and communicating
well than by using the most advanced forecasting technique or
model