Chapter 3 Demand

Forecasting
James Kaconco
Faculty of Technology Room 217
0772 653191
 Chapter Objectives
• By the end of the chapter students should be in position
to
1. Define Forecast and List fields in which forecasts
are applied
2. List elements of a good forecast
3. List and explain components of forecasts
4. List the principles of forecasting
5. Explain how to accurately collect data to use in
forecasting
6. Apply forecasting methods (qualitative and
quantitative)
7. Determine forecast accuracy
 Forecasting Defined & Areas of
Application
• Forecast: statement about the future (occurrence, timing and
magnitude of uncertain events).
– Managers of business firms attempt to predict how much of
their product will be demanded in the future
– Forecasts results in more accurate inventory and the smooth
operations of business organizations
– Forecasts provide information that can assist managers in
guiding future activities towards meeting organizational goals
• Forecasts are used to plan the utilization of a productive system
and focuses on:
– Plan the system (long-range plans) - used for location, capacity, and
new-product decisions
– Plan the use of the system (short-range plans) - such as those for
production-and-inventory control, labour levels, and cost controls–can
rely more on recent history
– Provide future goals (customer satisfaction, technological changes,
etc)
 Forecasting Objectives and Uses
• Forecast Objectives
– Better use of capacity (based on demand forecast, personnel
level required)
– Improve customer satisfaction (how much inventory to carry)
– Improved profitability (cost of raw materials)

• Areas of forecast applications

– New product development & Production
– Human Resource
– Marketing
– Finance & accounting
– IT and MIS
– Sports betting
 Elements of a good Forecast

1. Reliable: It should work consistently

2. Accurate: Degree of accuracy should be stated
3. Meaningful: It should be expressed in meaningful
units
4. Written: To guarantee use of the same information
and to make easier comparison to actual results.
5. Timely: Forecasting horizon must cover the time
necessary to implement possible changes
6. Easy to use: Users should be comfortable working
with the forecast
 Components of Forecasts
1. Time Frame: How far in the future to forecast
(Short, Medium and or Long)
2. Demand Behavior
a. Trend is a gradual, long-term, up-or-down movement of demand
(positive, negative, stagnant)
b. Cyclic is an up-and-down repetitive movement in demand (covers a
period of more than a year)
c. Seasonality is an up-and-down, repetitive movement within a trend
occurring periodically (follows seasons within a year, weeks in a
month, days in a week, hours in a day)
d. Random variations are movements that are not predictable and
follow no pattern (and thus are virtually unpredictable).
• Promotions
• Disasters
• Negative publicity
 Forecasting Principles
1. Forecasts are always wrong; forecasting is rarely
perfect (deviation is expected).
2. Every forecast should include an estimate of error.
Errors follow a normally distributed curve
3. Forecasts are more accurate for groups of products
than the forecast for individual items.
4. Forecast are more accurate for shorter than longer
time periods
5. Forecasting techniques assume that there is some
degree of stability in the system, and what happened
in the past will continue to happen in the future
6. Many types of forecasting models exist; each differ in
complexity and amount of data needed
 Forecasting Methods
• A variety of forecasting methods exist, and their
applicability is dependent on the:
– time frame of the forecast (i.e., how far in the future
we are forecasting),
– existence of patterns in the forecast (i.e., seasonal
trends, peak periods),
– Type of forecast (demand or technological)
– Data base (available or not available), and
– Methodology employed (qualitative or quantitative)
Note:
• Forecasting methods do not work accurately in all
situations; some work better than others.
 Forecasting Methods
• Qualitative methods – Subjective or Judgmental
methods
– Forecast generated subjectively or educated guesses
• based on experience and personal insights of forecaster
• Results may differ from individual to individual
– Used when there is need to make quick forecast
– Used when historical data is not available
– Techniques Include: executive opinions, sales-force
opinions, consumer surveys, and the Delphi method
• Quantitative methods – Objective methods:
– Forecasts generated through mathematical modeling and
heavily depend on historical data
– Techniques include: time series, regression, and casual
methods
 Quantitative or Objective Methods
• Time Series Models:
– A time series is a set of observations of a variable at
regular intervals over time.
– Time series mode: assumes information needed to
generate a forecast is contained in a time series of
data
– Assumes the future will follow same patterns as the
past (trend, cyclical, seasonal, random or
irregular)
• Causal Models or Associative Models
– Explores cause-and-effect relationships
– Uses leading indicators to predict the future
• E.g. housing starts and appliance sales
 Technological Forecasting
• Historical data can not be used (judgement options)
• Has become increasingly crucial factor for successful
competition in today's global business environment.
– New enhanced computer technology, new production methods,
and advanced machinery and equipment are constantly being
made available to companies.
• Technological Forecasting enable companies to
introduce more new products into the marketplace
faster than ever before.
– The companies that succeed do so by getting a technological
jump on their competitors through accurate prediction of future
• technology and its capabilities. What new products and services will be
technologically feasible, when they can be introduced, and what their
demand will be.
 Time Series Models
• Time series methods are statistical techniques that
make use of historical data accumulated over a period
of time.
• Naive:
• Simple Mean:
• Moving Average
• Weighted Moving Average
• Exponential smoothing
• Adjusted exponential smoothing
• Regression Model
• Seasonality model
 Naïve Method
• The forecast is equal to the actual value
observed during the last period – good for level
patterns
ORDERS
MONTH PER MONTH FORECAST
Jan 120 -
Feb 90 120
Mar 100 90
Apr 75 100
May 110 75
June 50 110
July 75 50
Aug 130 75
Sept 110 130
Oct 90 110
Nov - 90
Time Series: Simple Mean or
Average

n

i=1
Di
MAn =
n
where

n = number of periods in
the moving average
Di = demand in period i
Time Series Model: Moving Average:
MA (n)
– uses fixed several values during the recent
past to develop a forecast
• Fixed number of periods to consider in
determining the average is n
• The average moves while keeping (n) constant
• Each new demand value added pushes the
oldest data point out of the list
• Moving averages can smooth out fluctuations in
any data
• More responsive to a trend but still lags behind
actual data
Moving Average: MA(n)
n
 Di
i=1
MAn =
n
where

n = number of periods in
the moving average
Di = demand in period i
 3-month Moving Average

ORDERS MOVING
MONTH PER MONTH AVERAGE 3

Jan 120 – 
i=1
Di
Feb 90 – MA3 =
Mar 100 – 3
Apr 75 103.3
May 110 88.3 90 + 110 + 130
June 50 95.0
= 3
July 75 78.3
Aug 130 78.3
= 110 orders for Nov
Sept 110 85.0
Oct 90 105.0
Nov - 110.0
 5-month Moving Average
ORDERS MOVING
MONTH PER MONTH AVERAGE 5

Jan 120 – 
i=1
Di
Feb 90 – MA5 =
Mar 100 – 5
Apr 75 –
May 110 – 90 + 110 + 130+75+50
= 5
June 50 99.0
July 75 85.0
Aug 130 82.0 = 91 orders for Nov
Sept 110 88.0
Oct 90 95.0
Nov - 91.0
 Smoothing Effects
150 –

125 – 5-month

100 –
Orders

75 –

50 – 3-month

Actual
25 –

0– | | | | | | | | | | |
Jan Feb Mar Apr May June July Aug Sept Oct Nov
Month
 Time Series Models: Weighted
Moving Average: WMA(n)
• Previous demand data is weighted
• Fixed (n) periods are considered in forecast
• All weights must add to 100% or 1.00
e.g. Dt: 0.5, Dt-1: 0.3, Dt-2: 0.2 (weights add to 1.0)
• Weights are subjectively determined
• Allows emphasizing one period over others; above
indicates more weight on recent data (Dt = 0.5)
• Each new demand value drops the oldest data point &
adds a new observation
• Differs from the simple moving average that weighs all
periods equally - more responsive to trends
 Weighted Moving Average

• Adjusts moving average method to more closely

reflect data fluctuations
n
WMAn =  Wi Di
i=1
where
Wi = the weight for period i,
between 0 and 100
percent
 Wi = 1.00
 Weighted Moving Average
Example

MONTH WEIGHT DATA

August 17% 130
September 33% 110
October 50% 90
3
November Forecast WMA3 = 
i=1
Wi Di

= (0.50)(90) + (0.33)(110) + (0.17)(130)

= 103.4 orders
 Time Series: Exponential
Smoothing
• Method is an averaging method that weights
the most recent past data more strongly
than more distant past data.
– Reacts to both previous (demand and forecast)
data
 Need just three pieces of data to start:
 Last period’s forecast (Ft); initialization
 Last periods actual value (Dt)
– Smoothing coefficient, (α), is a number between 0
and 1 that enters multiplicatively into each
forecast but whose influence on demand declines
exponentially as the data become older.
Exponential Smoothing: Formula
and Initialization

Ft +1 = Dt + (1 - )Ft
where:
Ft +1 = forecast for next period
Dt = actual demand for present period
Ft = previously determined forecast for
present period
= weighting factor, smoothing constant
Effect of Smoothing Constant

 Range: 0.0  1.0

 And subjectively determined
If = 0.20, then Ft +1 = 0.20 Dt + 0.80 Ft

If = 0, then Ft +1 = 0 Dt + 1 Ft = Ft
Forecast does not reflect recent data

If = 1, then Ft +1 = 1 Dt + 0 Ft = Dt
Forecast based only on most recent data
 Exponential Smoothing (α=0.30)

PERIOD MONTH DEMAND F2 = D1 + (1 - )F1

1 Jan 37
= (0.30)(37) + (0.70)(37)
2 Feb 40
3 Mar 41 = 37
4 Apr 37
F3 = D2 + (1 - )F2
5 May 45
6 Jun 50 = (0.30)(40) + (0.70)(37)
7 Jul 43 = 37.9
8 Aug 47
9 Sep 56 F13 = D12 + (1 - )F12
10 Oct 52 = (0.30)(54) + (0.70)(50.84)
11 Nov 55
= 51.79
12 Dec 54

Copyright 2011 John Wiley & 12-31

Sons, Inc.
 Exponential Smoothing
FORECAST, Ft + 1
PERIOD MONTH DEMAND ( = 0.3) ( = 0.5)
1 Jan 37 – –
2 Feb 40 37.00 37.00
3 Mar 41 37.90 38.50
4 Apr 37 38.83 39.75
5 May 45 38.28 38.37
6 Jun 50 40.29 41.68
7 Jul 43 43.20 45.84
8 Aug 47 43.14 44.42
9 Sep 56 44.30 45.71
10 Oct 52 47.81 50.85
11 Nov 55 49.06 51.42
12 Dec 54 50.84 53.21
13 Jan – 51.79 53.61

Copyright 2011 John Wiley & 12-32

Sons, Inc.
 Exponential Smoothing
70 –

60 – Actual  = 0.50

50 –

40 –
Orders

 = 0.30
30 –

20 –

10 –

0– | | | | | | | | | | | | |
1 2 3 4 5 6 7 8 9 10 11 12 13
Month
Copyright 2011 John Wiley & 12-33
Sons, Inc.
 Time Series Models: Seasonality

• Calculate the total demand per season and for

all periods
• Calculate a seasonal index for each season
– Divide the actual demand of each season by
the total demand for all periods
• Forecast demand for the next year or period
using any of the time series model.
• Allocate the forecast to seasons using the
seasonality indices.
 Example: Seasonality
DEMAND (1000’S PER QUARTER)
YEAR 1 2 3 4 Total
2002 12.6 8.6 6.3 17.5 45.0
2003 14.1 10.3 7.5 18.2 50.1
2004 15.3 10.6 8.1 19.6 53.6
Total 42.0 29.5 21.9 55.3 148.7

D1 42.0 D3 21.9
S1 = = = 0.28 S3 = = = 0.15
D 148.7 D 148.7
D2 29.5 D4 55.3
S2 = = = 0.20 S4 = = = 0.37
D 148.7 D 148.7
 Seasonality Forecasting

For 2005; using regression linear model

y = 40.97 + 4.30x = 40.97 + 4.30(4) = 58.17

SF1 = (S1) (F5) = (0.28)(58.17) = 16.28
SF2 = (S2) (F5) = (0.20)(58.17) = 11.63
SF3 = (S3) (F5) = (0.15)(58.17) = 8.73
SF4 = (S4) (F5) = (0.37)(58.17) = 21.53
Other models can also be used to
forecast year 2005
 Time Series: Linear Regression
Model
• The model is used when the independent
variable is based on time(t)
• Often, leading indicators can help to predict
changes in future demand
– Sales of cards with calendar time,
– Purchase of decoder airtime with calendar time
• Regression models establish a cause-and-
effect relationship between independent and
dependent variables
• More than one independent variable may require
multiple regression modeling
 Regression and Correlation Method
• Simple regression expresses the relationship between a
dependent variable Y and a independent variable X in terms of the
slope and intercept of the line of best fit relating the two variables.
– The simple linear regression model takes the form Y = a + bX,
– Values for the slope b and intercept a are obtained by using the normal
equations
• Simple correlation expresses the degree or closeness of the
relationship between two variables
– correlation coefficient that provides direction and magnitude for the line of
best fit.
– Neither regression nor correlation gives proof of a cause-effect relationship.
 Linear Regression Trend
Formulae
xy - nxy
y = a + bx b =
x2 - nx2
a = y-bx
where
a = intercept where
b = slope of the line n = number of periods
x = time period
x
y = forecast for x = = mean of the x values
demand for period x n
y
y = n = mean of the y values
Linear Regression Trend Line
Linear Regression Problem: A maker of golf shirts has been tracking
the relationship between sales and advertising dollars. Use linear
regression to find out what sales might be if the company invested
$53,000 in advertising next year.

Sales $ Adv.$
XY X2 Y2
(Y) (X)
b
 XY  n XY
 X  nX
2 2

1 130 32 4,160 2,304 16,900

2 151 52 7,852 2,704 22,801

28202 447.25147.25
b  0.232
3 150 50 7,500 2,500 22,500 10533 447.25
2

4 158 55 8,690 3,025 24,964 a  Y  b X  147.25 0.23247.25

a  136.29
5 153.9 53
Y  a  bX  136.29 0.232X
Tot 589 189 28,202 10,533 87,165
Y  136.29  0.23253  148.59
Avg 147.3 47.3
 Correlation Coefficient (r)
– Measures the direction (positive, negative, zero) and
– Measures strength (-1 to +1) of the linear relationship between
two variables. The closer the r value is to +1.0 the better the
regression line fits the data points.
• r is positive if Y increases as X increases, and
• r is negative if Y decreases as X increases.
• r of zero indicates an absence of any relationship between the two
variables.
 Correlation Coefficient (r)
• Correlation coefficient (r):

– For better results sample size must be more than 30

• Coefficient of determination (r2 ) measures the amount of

variation in the dependent variable about its mean that is
explained by the independent variable.
– Values of (r2 ) close to 1.0 are desirable.
 Time Series: Causal Regression
Models
• They are used when the independent variable
is not based on time(t)
• Causal models establish a cause-and-effect
relationship between independent and
dependent variables
• Often, leading indicators can help to predict
changes in the dependent variable
– Promotion effect on sales
– Maintenance costs on machine life time
• multiple regression modeling is used if there
is more than one independent variables are
involved
 Measuring Forecast Error
• Forecast error is the deviation of the actual
from the forecast values.
– Errors will vary from plus to minus,
• over-forecasts = negative errors and
• under-forecasts = positive errors
– Errors tend to average out near zero if the forecast
is on target.
– Forecasts are never perfect
• Measuring forecast error:
et = Dt – Ft
– Errors help to determine reliability of forecasting
method
 Measuring Forecasting Accuracy
• Mean Absolute Deviation  actual  forecast
(MAD) MAD 
n
– measures the total error in a forecast
without regard to sign
• Cumulative Forecast Error
(CFE) CFE   actual  forecast
– Measures any bias in the forecast

• Mean Square Error (MSE)

– Penalizes larger errors  actual - forecast2

– Should be zero MSE 

n

• Tracking Signal CFE

– Measures if your model is working TS 
– Normally distributed MAD
 Measuring Forecasting Accuracy

• Tracking signals are one way of monitoring

how well a forecast is predicting actual values.

• Action limits for tracking signals commonly range from

three to eight.
• Corrective action may be required when tracking signal
gets out of the range.
 Control Charts & Forecast Error
• Control charts are a second way of monitoring forecast error.
• Variations of actual from forecast (or average) values are quantified
in terms of the estimated standard deviation of forecast SF .

• Control limits are then set, at two or three standard deviations

away from the forecast average X or the 2SF or 3SF
– Control limits are used as maximum acceptable limits for forecast error. Control
limits are based on individual forecast values.
• Forecast errors are normally distributed around the forecast
average.