Professional Documents
Culture Documents
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Chapter 6: Forecasting
1. Introduction
Forecasting is vital to every business organization and for every significant
management decision. Forecasting is the basis of corporate long-run planning. In
the functional areas of finance and accounting, forecasts provide the basis for
budgetary planning and cost control. Marketing relies on sales forecasting to plan
new products, compensate sales personnel, and make other key decisions.
Production and operations personnel use forecasts to make periodic decisions
involving process selection, capacity planning, and facility layout, as well as for
continual decisions about production planning scheduling, and inventory.
Bear in mind that a perfect forecast is usually impossible. Too many factors in the
business environment cannot be predicted with certainty. Therefore, rather than
search for the perfect forecast, it is far more important to establish the practice of
continual review of forecasts and to learn to live with inaccurate forecasts. When
forecasting, a good strategy is to use two or three methods and look at them for the
commonsense view.
2. Types of forecasting
Forecasting can be classified into four basic types:
Qualitative: Qualitative techniques are subjective or judgmental and are based on
estimates and opinions.
Time series analysis: This is based on the idea that data relating to past demand
can be used to predict future demand.
Causal relationships: Causal forecasting, which we discuss using the linear
regression technique, assumes that demand is related to some underlying factor or
factors in the environment.
Simulation: Simulation models allow the forecaster to run through a range of
assumptions about the condition of the forecast.
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
about existing products, which competitive products within a particular class are
preferred, and so on. The data collection methods are primarily surveys and
interviews.
Historical analogy: In trying to forecast demand for a new product, an ideal situation
would be where an existing product or generic product could be used as a model.
There are many ways to classify such analogies--for example, complementary
products, substitutable or competitive products, and products as a function of
income. If you buy a CD through the mail, you will receive more mail about new CDs
and CD players. Another example would be toasters and coffee pots. A firm that
already produces toasters and wants to produce coffee pots could use the toaster
history as a likely growth model.
Panel consensus: In a panel consensus, the idea that two heads are better than one
is extrapolated to the idea that a panel of people from a variety of positions can
develop a more reliable forecast than a narrower group. Panel forecasts are
developed through open meetings with free exchange of ideas from all levels of
management and individuals. The difficulty with this open style is that lower
employee levels are intimidated by higher levels of management. For example, a
salesperson in a particular product line may have a good estimate of future product
demand but may not speak up to refute a much different estimate given by the vice
president of marketing. The Delphi technique overcomes this problem.
Delphi method: this method conceals the identity of the individuals participating in
the study. Everyone has the same weight. Procedurally, a moderator creates a
questionnaire and distributes it to participants. Their responses are summed and
given back to the entire group along with a new set of questions. This technique can
usually achieve satisfactory results in three rounds.
2/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Actual
Demand
800
1400
1000
1500
1500
1300
1800
1700
1300
1700
1700
1500
2300
2300
2000
3-week
9-week
3/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Although it is important to select the best period for the moving average, there are
several conflicting effects of different period lengths. The longer the moving average
period, the more the random elements are smoothed. But if there is a trend in the
data either increasing or decreasing the moving average has the adverse
characteristic of lagging the trend. Therefore, for a shorter time span, there is a
closer following of the trend. Conversely, a longer time span gives a smoother
response but lags the trend.
The main disadvantage with this method is that all individual elements must be
carried as data because a new forecast period involves adding new data and
dropping the earliest data. The amount of data involved is significant.
Another shortcoming of this method is it lags behind the trend. Consider a demand
process in which there is a definite trend as follows.
Period
Actual
demand
3-period avg
6-period avg
10
11
12
8
4
10
6
12
8
14
10
7
16
12
9
18
14
11
20
16
13
22
18
15
24
20
17
Notice that both the 3-period and 6-period moving average forecasts lag behind the
trend, and that the forecast with a smaller N value follows the actual demand more
closely.
4.2.2. Weighted moving average
Whereas the simple moving average gives equal weight to each component of the
moving average database, a weighted moving average allows any weights to be
placed on each element, providing, of course, that the sum of all weights equals 1.
For example, a department store may find that in a four-month period, the best
forecast is derived by using 40 percent of the actual sales for the most recent
month, 30 percent of two months ago, 20 percent of three months ago, and 10
percent of four months ago. If actual sales experience was
month 1
100
month 2
90
month 3
105
month 4
95
month 5
?
4/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Choosing weights: experience and trial and error are the simplest ways to choose
weights. As a general rule, the most recent past is the most important indicator of
what to expect in the future, and therefore, it should get higher weighting. However,
if the data are seasonal, for example, weights should be established accordingly.
Bathing suit sales in July of last year should be weighted more heavily than bathing
suit sales in December.
Example 2:
In Atlanta, the number of daily calls for repair of Speedy copy machines in 8 days
has been recorded as follows:
actual
demand
day
1
2
3
4
5
6
7
8
3-day moving
avg
Forecast
error
weighted moving
avg
Forecast
error
92
127
103
165
132
111
174
94
a. Prepare a 3-period moving average forecast. What is the error on each day?
b. Prepare a 3-period weighted moving average forecast with w1=0.2, w2=0.3, and
w3=0.5(most recent data carries heaviest weight). What is the error on each
day?
c. Which of the two forecasts is better? (use MAD to judge)
5/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
6/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
The following is some supplementary info for you to understand why we use the
term exponential. You can skip it if you are not interested.
Notice that Ft-1 = Ft-2 + (At-2 Ft-2), substituting into the above equation yields
Ft = At-1 + (1- ) At-2 + (1- )2 Ft-2
We can substitute for Ft-2 in the same fashion. If we continue in this way we obtain
the infinite expansion for Ft:
Ft = (1 ) i At i 1
i =0
Hence, exponential smoothing applies a declining set of weights to all past data. In
fact, we can fit the continuous exponential curve exp(-i) to these weights, which is
why the method is called exponential smoothing. The smoothing constant plays
essentially the same role here as the value of N does in moving averages. If is
large, more weight is placed on the current observation of demand and less weight
on past observations, which results in forecasts that will react quickly to changes in
the demand pattern but may have much greater variation from period to period.
7/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
8/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
a = y bx
b=
xy nx. y
x nx
2
Example 5:
A firms sales for a product line during the 12 quarters of the past three years were
as follows:
Quarter
1
2
3
4
5
6
Sales
600
1550
1500
1500
2400
3100
Quarter
7
8
9
10
11
12
Sales
2600
2900
3800
4500
4000
4900
The firm wants to forecast each quarter of the fourth yearthat is, quarters 13
through 16.
Solution:
x
xy
600
1550
1500
1500
2400
3100
2600
2900
3800
10
4500
11
4000
12
4900
x2
9/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
OPIM201
BUSINESS PROCESSES
-------------------------------------------------------------------------------------------------------------------------------------------Quarter
1
2
3
4
Year 1
45
335
520
100
Year 2
70
370
590
170
Year 3
100
585
830
285
Year 4
100
725
1160
215
The manager wants to forecast customer demand for each quarter of year 5, based
on her estimate of total year 5 demand of 2600 customers.
Solution:
Overall Avg quarterly sales in past years =
Avg quarterly sales for next year =
avg past sales
seasonal
factor
Quarter 1
Quarter 2
Quarter 3
Quarter 4
Example 7:
11/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Use the same data given in example 5, but now we consider the seasonal effect.
Suppose the seasonal factor is given by the average for the same quarters in the 3year period divided by the general average for all 12 quarters. What are the
forecasts for the quarters 13 through 16?
Solution:
Overall quarterly avg. =
avg
seasonal factor
Spring
Summer
Fall
Winter
Quarter
Sales
600
1,550
1,500
1,500
2,400
3,100
2,600
2,900
3,800
10
4,500
11
4,000
12
4,900
deseasonalized
demand
Quarter
Trend-based
forecast
Final forecast
13
14
15
16
12/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
1 n
ei
n i =1
MSE =
1 n 2
ei
n i =1
Note that the MSE is similar to the variance of a random sample. The MAD is often
the preferred method of measuring the forecast error because it does not require
squaring. Furthermore, when forecast errors are normally distributed, as is generally
assumed, an estimate of the standard deviation of the forecast error, e, is given by
1.25 times the MAD.
Another measure of forecast accuracy is known as the mean absolute percentage
1 n e
error (MAPE) and is given by MAPE = i * 100 . It is independent of the
n i =1 Ai
magnitude of the values of demand.
4.5.2. Criteria for selecting time-series methods
The criteria to use in making forecast method and parameter choices include
Minimizing bias
Minimizing MAD or MSE
Meeting managerial expectations of changes in the components of demand
Minimizing the forecast error last period.
However, managers recognize that the best technique in explaining the past data is
not necessarily the best to predict the future. For this reason, some analysts prefer
to use a holdout set as a final test. To do so, they set aside some of the more recent
periods from the time series, and use only the earlier time periods to develop and
test different model. Once the final models have been selected in the first phase,
then they are tested again with the holdout set.
13/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Permits
18
15
12
10
20
28
35
30
20
Sales
13000
12000
11000
10000
14000
16000
19000
17000
13000
Suppose that there are 25 permits for houses to be built in 2000. What is the
forecast for sales in 2000?
6. Concluding remarks
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
7. Exercises:
Question 1:
Sunrise Baking Company markets doughnuts through a chain of food stores. It has been
experiencing over- and under-production because of forecasting errors. The following data are
its demand in dozens of doughnuts for the past four weeks. Doughnuts are made for the
following day; for example, Sundays doughnut production is for Mondays sales, Mondays
production is for Tuesdays sales, and so forth. The bakery is closed Saturday, so Fridays
production must satisfy demand for both Saturday and Sunday.
4 weeks ago
Monday
2200
Tuesday
2000
Wednesday 2300
Thursday
1800
Friday
1900
Saturday
Sunday
2800
3 weeks ago
2400
2100
2400
1900
1800
2 weeks ago
2300
2200
2300
1800
2100
last week
2400
2200
2500
2000
2000
2700
3000
2900
Question 2:
Here are quarterly data for the past two years. From these data, prepare a forecast for the
upcoming year using decomposition.
Period
1
2
3
4
5
6
7
8
Actual demand
300
540
885
580
416
760
1191
760
15/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Question 3:
The demand for Krispee Crunchies, a favorite breakfast cereal of people born in the 1940s, is
experiencing a decline. The company wants to monitor demand for this product closely as it
nears the end of its life cycle. The trend-adjusted exponential smoothing method is used with
alpha = 0.1 and delta=0.2. At the end of December, the January estimate for the average
number of cases sold per month, FJan, was 900000 and the trend, TJan, was -50000 per month.
The following table shows the actual sales history for Jan, Feb, and Mar. Generate forecast for
Feb, Mar, and Apr.
Month
Jan
Feb
Mar
Sales
890,000
800,000
825,000
Question 4:
The Northville Post Office experiences a seasonal pattern of daily mail volume every week. The
following data for two representative weeks are expressed in thousands of pieces of mail:
Day
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
total
week 1
5
20
30
35
49
70
15
224
week 2
8
15
32
30
45
70
10
210
16/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Answer Key
Question 1:
A.
Question 2:
Period
1
2
3
4
5
6
7
8
Avg
Actual demand
300
540
885
580
416
760
1191
760
679
period avg
358
650
1038
670
seasonal facto
0.527
0.957
1.529
0.987
0.5272
0.9573
1.5287
0.9867
deseasonalized demand
568.99
564.09
578.92
587.79
789.01
793.91
779.08
770.21
Run a regression in Excel using deseasonalized demand, we obtain the parameter values: a = 500.6,
b=39.64
Therefore, we have:
seasonal factor
0.527
0.957
1.529
0.987
final forecast
452.0
858.7
1431.9
963.3
17/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
Question 3:
Feb: 804800
Mar: 755024
Apr: 714125
Question 4:
Day
week 1
Sunday
Monday
Tuesday
Wednesday
Thursday
Friday
Saturday
total
5
20
30
35
49
70
15
224
week
2
8
15
32
30
45
70
10
210
daily
avg
6.5
17.5
31
32.5
47
70
12.5
31
seasonal factor
0.210
0.565
1.000
1.048
1.516
2.258
0.403
forecast
6,889
18,548
32,857
34,447
49,816
74,194
13,249
18/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
8. Appendix
19/20
OPIM201
BUSINESS PROCESSES
--------------------------------------------------------------------------------------------------------------------------------------------
20/20