You are on page 1of 38

Operations

Management
MS29P
Forecasting
D. Anthony Chevers
1

Lecture #4 - Forecasting

Definition
Planning horizon
Forecasting techniques & comparison
Simple moving average
Weighted moving average
Exponential smoothing
Forecast errors
Regression analysis
Exercises
2

Learning Objectives
When you complete this chapter
you should be able to :
1. Understand the three time
horizons and which models apply
for each use
2. Explain when to use each of the
three qualitative models
3. Apply moving average,
exponential smoothing, and
regression analysis

Learning Objectives
When you complete this chapter
you should be able to :
4. Compute two measures of
forecast accuracy
5. Compare and contrast each
technique taught

Forecasting - Defined

Forecasting is the prediction/estimation of


future activities. Types of forecast Economic,

Technological and Demand


The first step in planning is forecasting, or estimating
the future demand for products and services and the
resources necessary to produce these outputs. E.g. Rising
Star finals (2005); Craft Village New Kgn for World Cup Cricket 2007 -closure & Diana Ross (Jazz Festival,
Jan 2008); Girl you have Intimate potential Gregory Isaacs (2008)

Operations managers need long range forecasts to


make strategic decisions about products, processes
and facilities.
Operations managers need short range forecasts to
assist them in making decisions about production
issues that span only the next few weeks.
5

Planning Horizon,
Tasks &
Responsibilities

Top Executives Long range plans, over 1 year


R & D
New product line
Capital expenditure
Facility location/expansion
Operations Managers Middle term plans, 3-18 months
Sales planning
Production planning
Setting inventory levels
Supervisors Short term plans, up to 3 months
Job assignments
Ordering
Job scheduling/Dispatching
6

Forecasting
Techniques
Qualitative Methods

Time Series Methods

Market research, Delphi, Panel consensus,


Grass roots & Historical analogy
Moving average, Exponential smoothing,
Box Jenkins & Trend projections

Causal Methods

Regression analysis, Econometric models,


Input/Output models, Life-Cycle analysis &
Simulation
8

Analysis of some
Forecasting Techniques
..

Market Research
Accuracy = Excellent
Cost = High
Availability-historical data = None
Availability-competent men = Yes
Time needed for analysis = Long
Forecast time horizon = Long
Moving Average
Accuracy = Good
Cost = Low
Availability-historical data = Yes
Availability-competent men = Partial
Time needed for analysis = Short
Forecast time horizon = Short

Analysis of some
Forecasting Techniques

Exponential Smoothing [Tutorial]

Accuracy = ?
.#2
Cost = ?

Availability historical data = ?


Availability competent men = ?
Time needed for analysis = ?
Forecast time horizon = ?
Exercise: Rank exponential smoothing in terms of factors above

Regression Analysis
Accuracy = Very good
Cost = Medium
Availability-historical data = Yes
Availability-competent men = Yes
Time needed for analysis = Medium
Forecast time horizon = Long
10

Equations - Simplified

SMA = Demand / n
WMA = (Demand x Weight) / Weight
Exp. Smooth: Ft = Ft-1 + (At-1 Ft-1)
MAD = (l Forecast error l) / n
MSE = (Forecast error)2 / n
= a + bX
Ya = y bx

XY nXY
b
X error
nX = Actual - Forecast
Forecast

Simple Moving
Average
January

Actual tables
[6 Months, Moving Average65
(M.A.)]

February
March
April
May
June

Forecast??

July ????

sold
93
85
105
71
115
534 = Total
534/6 = 89 = M.A.
12

Rearranged Demand
Schedule
[6 Months M.A.]
65
115

71
85
93
105
115

89 = Julys Avg.

105
93
85
71
65

89 = Julys Avg.
13

Simple Moving
Averages
Moving Average (MA) = demand in
previous n periods / n

Where n is the number of periods in the M.A.

Example 1 Phone Sales

Month
Jan
Feb
Mar
Apr
May

Actual Sales
3-month MA
100
120
130
160 (100+120+130)/3=116.7
190 (120+130+160)/3=136.7
14

Weighted Moving Averages


Weighted MA = [(weight for period n) (demand in period n)] /
(WMA)
weights

Example 2 Phone Sales


Weights applied
Period
{3}
Last month (most recent)
[Given]
{2}
Two month ago
[Given]
{1}
Three month ago
[Given]
[6]
Month
Jan 100
Feb120
Mar
Apr?
May?

Sum of weights
Actual Sales

130
160
190

[Calculated]

3-month WMA

[(3x130)+(2x120)+(100)]/6=121.7
[(3x160)+(2x130)+(120)]/6=143.3

15

Details in Different
Format
[WMA = [Weight x Demand] / Weight

16

Exponential
Smoothing

Exponential smoothing is a sophisticated weighted movingaverage forecasting method that is still fairly easy to use
It is ideally suited to short-run forecasting for inventory control
The basic formula can be shown as:
New forecast =last periods forecast + @(last periods
actual demand last periods forecast) --(1)
Equation (1) can be rewritten mathematically as:
F = F
--(2)
t
t-1 + @(At-1 Ft-1)

Where Ft = new forecast or Forecast of period t


Ft-1 = previous forecast or Forecast before t
@ = smoothing constant (0 < @ < 1)
At-1 = previous periods actual demand

17

Example 3

In January, a car dealer predicted a February


demand for 142 Ford pick-ups. Actual February
demand was 153. Using a smoothing constant
chosen by management, of @ = 0.20, we can
forecast the March demand using the exponential
smoothing model. Substituting into the formula,
we obtain

FMarch = FFeb + @ (AFeb FFeb)


New forecast (for March demand) = 142 + 0.2(153 142)
= 142 + 0.2(11)
= 142 + 2.2
= 144.2

Thus, the demand forecast for Ford Pick-ups in


March is rounded to 144
18

Example 4

The Port of Bustamante has unloaded large quantities of


grain from ships during the past four quarters. The Ports
operations manager wants to test the use of exponential
smoothing to see how well the techniques works in
predicting tonnage unloaded. He guesses that the forecast
of grain loaded in the first quarter was 175 tons. Two
values of @ are examined: @=0.10 and @=0.50. The
following table shows the detailed calculations for @=0.10
only:

Qtr
1
2
3
4

Actual tons Forecast @=0.1


Forecast @=0.5
180 175
175
168 175+0.1(180-175)=175.5 178
159 175.5+0.1(168-175.5)=174.75
173
175 174.7+0.1(159-174.7)=173.1
166

19

Monitoring Accuracy

Accuracy of a forecasting approach needs


to be monitored to assess the confidence
you can have in its forecasts
The forecast error is defined as

Forecast error = Actual demand - Forecast

Accuracy can be measured in several


ways

Mean absolute deviation (MAD)


Mean squared error (MSE)
20

Forecast Error
MAD
[n = 4 (#of Periods)]
(Deviations)/n]

[MAD =

21

Forecast Error MSE


[n = 4 (# of periods)]

[MSE =

(Error)2/n]

22

Regression Analysis

Causal forecasting models usually consider several


variables that are related to the quantity being
predicted
This approach is more powerful than the time-series
methods that use only the historical values for the
forecasting variable
For example, the sales of PCs might be related to
advertising budget, the price charged, competitors
prices, promotional strategies, the economy,
disposable income or unemployment rates.
In this case, PC sales would be called the dependent
variable (Y) and the other variables would be called
independent variables (X).
We will use Y with one other variable (X)
The most common quantitative casual forecasting
model is linear-regression analysis
23

Modular Company
Sales
Vs Local
Payroll
Y ($)
X ($)

(Mths)

25
40
60
35
90
120
100
50
55
70
??

15
25
40
20
60
80
70
20
25
40
[45] Projection
24

Regression Equations

Y = a + bx .(1)
b = [nxy - xy]/[nx2 (x)2]
a = Y - bX .(3)
Y = y/n
X = x/n

.(2)

Note: means summation, n is # of periods &


bold
Y and X are averages.

25

b = [nxy - xy]/[nx2 (x)2] .(2)

26

Modular Solution
n = 10
b = 1.29
a = 13.55
Y = 13.55 + 1.29x
If X is 45, then Y = 71.6

27

Tutorial Questions

#1

28

Tutorial Questions

#2

29

Tutorial Questions

#3

30

Solution Tutorial #1
Forecast used for new product planning, capital expenditure, facility location or
expansion and research and development are typically called:

long range forecast


medium range forecast
short range forecast
all of the above
none of the above

Three pieces of data are needed for exponential smoothing. They are: an alpha
value, actual sales for last period and forecasted sales for last period
Long range forecasting often includes finding independent and dependent
variables and using a statistical technique know as regression analysis
Rank exponential smoothing forecasting technique in terms of the following;
(a) accuracy [Good] (b) cost [Low]
(c ) availability of historical data [Yes]
(d) availability of skill men [Partial]
(e) time for analysis [Short]
(f) time horizon [Short]
5. For the data below, develop a 3-month moving average forecast
Automobile
Automobile
Month
Battery Sales
Month
Battery Sales
January 20
July
17 [(14+13+16)/3] = 14.3
February 21
August
18 [(13+16+17)/3] = 15.3
March
15
September
20 [(16+17+18)/3] = 17.0
April
14 [(20+21+15)/3] = 18.7
October 20 [(17+18+20)/3] = 18.3
May
13 [(21+15+14)/3] = 16.7
November
21
[(18+20+20)/3] = 19.3
June
16 [(15+14+13)/3] = 14.0
December
23
[(20+20+21)/3] = 20.3
31

Solution Tutorial #2

Given the following data, develop a 3-year moving average forecast of


demand.
Year
8
Demand
13
Forecast
11.0

1
9
7
9

2
10
9
11

3
11
5
7

11.0

11.3

11.0

13

12

7.0

7.7

9.0

10.0

7.
For the demand data in Table below, calculate the four months
moving average for June 2004 and Sept 2004.
Time period 2004
Actual Calculator demand (units)
4
Months M.A.
Jan
10,000
Feb
12,000
Mar
15,000
Apr
13,000
May
25,000
June
20,000 (12+15+13+25)/4 = 16,250
July
18,000
Aug
14,000
Sept
16,000 (25+20+18+14)/4 = 19,250
Oct
15,000
Nov
9,000
Dec
12,000

32

Solution Tutorial #8

{20.34/8}

[Simple & Weighted Moving Averages]

33

Solution Tutorial #9

34

Solution Tutorial #10

35

Solution Tutorial #11

36

Solution Tutorial #12

37

NEXT LECTURE:
Location & Layout
Dr. D. Anthony Chevers
delroy.chevers@uwimona.edu.jm
DOMS, Room #28
38

You might also like