5

5.1 5.6
5.2 5.7
5.3
5.4
5.5

Forecasts are essential for the smooth operations of business organizations. They provide information
that can assist managers in guiding future activities toward organizational goals.

Forecasts are estimates of the occurrence, timing, or magnitude of uncertain future events. Forecasts
are essential for the smooth operations of business organizations. They provide information that can
assist managers in guiding future activities toward organizational goals.
Operations managers are primarily concerned with forecasts of demand—which are often
made by (or in conjunction with) marketing. However, managers also use forecasts to estimate raw
material prices, plan for appropriate levels of personnel, help decide how much inventory to carry,
and a host of other activities. This results in better use of capacity, more responsive service to
customers, and improved profitability.

Forecasting activities are a function of (1) the type of forecast (e.g., demand, technological), (2) the
time horizon (short, medium, or long range), (3) the database available, and (4) the methodology
employed (qualitative or quantitative). Forecasts of demand are based primarily on non-random
trends and relationships, with an allowance for random components. Forecasts for groups of products
tend to be more accurate than those for single products, and short-term forecasts are more accurate
than long-term forecasts (greater than five years). Quantification also enhances the objectivity and
precision of a forecast.
There are numerous methods to forecasting depending on the need of the decision-maker. These
can be categorized in two ways:
1. Opinion and Judgmental Methods or Qualitative Methods.
2. Time Series or Quantitative Forecasting Methods.

Some opinion and judgment forecasts are largely intuitive, whereas others integrate data and perhaps
even mathematical or statistical techniques. Judgmental forecasts often consist of (1) forecasts by
individual sales people, (2) Forecasts by division or product-line managers, and (3) combined estimates
of the two. Historical analogy relies on comparisons; Delphi relies on the best method from a
group of forecasts. All these methods can incorporate experiences and personal insights. However,
results may differ from one individual to the next and they are not all amenable to analysis. So there
may be little basis for improvement over time.

A time series is a set of observations of a variable at regular intervals over time. In decomposition
analysis, the components of a time series are generally classified as trend T, cyclical C, seasonal S,
and random or irregular R. (Note: Autocorrelation effects are sometimes included as an additional
factor.)
Time series are tabulated or graphed to show the nature of the time dependence. The forecast
value (Ye) is commonly expressed as a multiplicative or additive function of its components; examples
here will be based upon the commonly used multiplicative model.
Yc = T. S. C. R multiplicative model (5.1)
Yc = T + S + C + R additive model (5.2)
where T is Trend, S is Seasonal, C is Cyclical, and R is Random components of a series.
Trend is a gradual long-term directional movement in the data (growth or decline).
Seasonal effects are similar variations occurring during corresponding periods, e.g., December
retail sales. Seasonal can be quarterly, monthly, weekly, daily, or even hourly indexes.
Cyclical factors are the long-term swings about the trend line. They are often associated with
business cycles and may extend out to several years in length.
Random component are sporadic (unpredictable) effects due to chance and unusual
occurrences. They are the residual after the trend, cyclical, and seasonal variations are removed.
SUMMARY OF FORECASTING METHODS
M e t hod Description Time Relative
horizon cost

Sales force composites Estimates from field sales people are aggregated SR-MR L-M
Executive opinion (and/or panels) Marketing, finance, and production managers jointly SR-LR L-M
prepare forecast
 Field sales and product-line management Estimates from regional sales people are reconciled MR M
with national projections from product-line managers
Historical analogy Forecast from comparison with similar product SR-LR L-M
previously introduced
Delphi Experts answer a series of questions (anonymously), LR M-H
receive feedback, and revise estimates
Market surveys Questionnaires/interviews for data to learn about MR-LR H
consumer behaviour

Naive Forecast equals latest value or latest plus or minus SR L

some percentage
Moving average Forecast is average of most recent periods SR L
(can also be weighted)
Trend projection Forecast is linear, exponential, or other projection MR-LR L
of past trend
Decomposition Time series is divided into trend, seasonal, cyclical, SR-LR L
and random components
Exponential smoothing Forecast is an exponentially weighted moving average, SR L
where latest values carry most weight
Box-Jenkins A time-series-regression model is proposed, statistically MR-LR M-H
tested, modified, and retested until satisfactory

Regression and correlation Use one or more associate variables to forecast via SR-MR M-H
(and leading indicators) a least-squares equation (regression) or via a close
association (correlation) with an explanatory variable
Econometric Use simultaneous solution of multiple regression SR-LR H
equations that relate to broad range of economic activity

Naive Forecast equals latest value or latest plus or SR L

minus some percentage
Moving average Forecast is average of most recent periods SR L
(can also be weighted)
Trend projection Forecast is linear, exponential, or other projection of MR-LR L
past trend
Decomposition Time series is divided into trend, seasonal, cyclical, SR-LR L
and random components
Exponential smoothing Forecast is an exponentially weighted moving SR L
average, where latest values carry most weight
Box-Jenkins A time-series-regression model is proposed, MR-LR M-H
statistically tested, modified, and retested until
satisfactory

Regression and correlation Use one or more associate variables to forecast via SR-MR M-H
(and leading indicators) a least-squares equation (regression) or via a close
association (correlation) with an explanatory variable
Econometric Use simultaneous solution of multiple regression SR-LR H
equations that relate to broad range of economic
activity
Key: L = low, M = medium, H = high, SR = short range, MR = medium range, LR = long range.
FORECASTING PROCEDURE FOR USING TIME SERIES
Following are the steps in time series forecasting:
1. Plot historical data to confirm relationship (e.g., linear, exponential).
2. Develop a trend equation (T) to describe the data.
3. Develop a seasonal index (SI, e.g., monthly index values).
4. Project trend into the future (e.g., monthly trend values).
5. Multiply trend values by corresponding seasonal index values.
6. Modify projected values by any knowledge of:
(C) Cyclical business conditions,
(R) Anticipated irregular effects.
Trend: Three methods for describing trend are: (1) Moving average, (2) Hand fitting, and
(3) Least squares.
1. MOVING AVERAGE
A centered moving average (MA) is obtained by summing and averaging the values from a given
number of periods repetitively, each time deleting the oldest value and adding a new value. Moving
averages can smooth out fluctuations in any data, while preserving the general pattern of the data
(longer averages result in more smoothing). However, they do not yield a forecasting equation, nor
do they generate values for the ends of the data series.

x
MA = Number of Period

A weighted moving average (MAw) allows some values to be emphasized by varying the weights
assigned to each component of the average. Weights can be either percentages or a real number.

(Wt ) X
MA wt =
Wt

I LLUSTRATION 1: Shipments (in tons) of welded tube by an aluminum producer are shown
below:

Year 1 2 3 4 5 6 7 8 9 10 11

Tons 2 3 6 10 8 7 12 14 14 18 19

(a) Graph the data, and comment on the relationship. (b) Compute a 3-year moving average,
plot it as a dotted line, and use it to forecast shipments in year 12. (c) Using a weight of 3 for the
most recent data, 2 for the next, and 1 for the oldest, forecast shipments in year 12.
 Demand

20
18 19
18
16
14
13 14
12 14
10 10

8 8
7
6 6
4
3
2 2
0
1 2 3 4 5 6 7 8 9 10 11
Years

Table 5.1 3-year moving average

Year Shipments (tonns) 3-year moving total 3-year moving average

1 2 - -
2 3 11 3.7
3 6 19 6.3
4 10 24 8.0
5 8 25 8.3
6 7 27 9.0
7 13 33 11.0
8 14 40 13.3
9 14 46 15.3
10 18 51 17.0
11 19 - -

(a) The data points appear relatively linear. (b) See Table 5.1 for computations and Fig. 5.2 for plot
of the MA. The MA forecast for year 12 would be that of the latest average, 17.0 tons.

(Wt ) X (1) (14) (2) (18) (3) (19)

(c) MA wt = 17.8 tons
Wt 1 2 3
2. HAND FITTING
A hand fit or freehand curve is simply a plot of a representative line that (subjectively) seems to best
fit the data points. For linear data, the forecasting equation will be of the form:
Yc = a + b (X) (signature)
where Yc is the trend value, a is the intercept (where line crosses the vertical axis), b is the slope
(the rise, y, divided by the run, x), and X is the time value (years, quarters, etc.). The “signature”
identifies the point in time when X = 0, as well as the X and Y units.
I LLUSTRATION 2: (a) Use a hand fit line to “develop a forecasting equation for the data
in Fig. 5.2. State the equation, complete with signature. (b) Use your equation to forecast tube
shipments for year 12.
(a) Select points some distance apart. A straight line connecting the values for years 3 and 8
might be a good
Free hand representation of the data. From this we can determine the slope and intercept:
Y Y2 Y1 14 6
Slope : b = 1.6 tons
X X 2 X1 8 3
Intercept: a = 0.5 tons (Note: This is the estimated Y value at X = 0 from graph.)
Equation: YC = 0.5 + 1.6X (Yr 0 = 0, X = yrs, Y = tons)
(b) For year 12: Yc =0.5 + 1.6(12) = 19.7 tons.

3. LEAST SQUARES
Least squares are a mathematical technique of fitting a trend to data points. The resulting line of
best fit has the following properties: (1) the summation of all vertical deviations about it is zero,
(2) the summation of all vertical deviations squared is a minimum, and (3) the line goes through the
means X and Y. For linear equations, the line of best fit is found by the simultaneous solution for a
and b of the following two normal equations:
Y = na b X
2
XY = a X b X
The above equations can be used in the form shown above and are used in that form for
regression. However, with time series, the data can also be coded so that X 0 . Two terms
then dropout, and the equations are simplified to:
Y
Y = na a
n

XY
XY = b X2 b .
X2
To code the time series data, designate the center of the time span as X = 0 and let each
successive period be ±1 more unit away. (For an even number of periods, use values of ±0.5, 1.5,
2.5, etc.).
I LLUSTRATION 3: Use the least square method to develop a linear trend equation for the
data from illustration 1. State the equation and forecast a trend value for year 16.
 Table 5.2

Year X year coded Y shipments (tons) XY X2

1 –S 2 – 10 25
2 –4 3 – 12 16
3 –3 6 – 18 9
4 –2 10 – 20 4
5 –1 8 –8 1
6 0 7 0 0
7 1 12 12 1
8 2 14 28 4
9 3 14 42 9
10 4 18 72 16
11 5 19 95 25

0 113 181 110

we have:
Y 113 XY 181
a = 10.30 b 2
1.6
n 11 X 110
The forecasting equation is of the form Y = a + bX.
Y = 10.3 + 1.6 X (year 6 = 0, X = years, Y = tons).
Seasonal indexes: A seasonal index (SI) is a ratio that relates a recurring seasonal
variation to the corresponding trend value at the given time. In the ratio-to-moving average
method of calculation monthly (or quarterly) data are typically used to compute a 12-month (or
4-quarter) moving average.
(This dampens out all seasonal fluctuations.) Actual monthly (or quarterly) values are then
divided by the moving average value centered on the actual month. In the ratio-to-trend
method, the actual values are divided by the trend value centered on the actual month. The
ratios obtained for several of the same months (or quarters) are then averaged to obtain the
seasonal index values. The indexes can be used to obtain seasonalized forecast values, Ysz (or
to deseasonalize actual data). Ysz = (SI) Yc.
I LLUSTRATION 4: Snowsport International has experienced low snowboard sales in July,
as shown in Table 5.3. Using the ratio-to-trend values, calculate a seasonal index value for
July and explain its meaning.
 Table 5.3

Yr 5 Yr 6 Yr 7 Yr 8 Yr 9 Yr 10 Yr 11 Yr l2
July actual sales 22 30 18 26 45 36 40
July trend value, 170 190 210 230 250 270 290

Ratio (actual trend) 0.13 0.16 0.09 0.11 0.18 0.13 0.14

Total = 0.94.
(a) A third row has been added to Table 5.3 to show the ratio of actual to trend values for
July. Using a simple average, the July index is SI July = 0.94 7 = 0.13. This means that July
is typically only 13 per cent of the trend value for July in any given year. Winter months are
likely quite high.
I LLUSTRATION 5: The forecasting equation for the previous example, centered in July of
year 4 with X units in months, was Yc = 1800 + 20X (July 15, Yr 4 = 0, X = mo, Y = units/yr).
* Use this equation and the July seasonal index of 0.13 to compute (a) the trend (deseasonalized)
value for July of year 12 and (b) the forecast of actual (seasonalized) snowboard sales in July
of year 12.
(a) July of year 12 is (8)(12) = 96 months away from July of year 4, so the/trend value is:
Yc = 1800 + 20(96) = 3,240 units/yr or 3,240 units/yr 12 mo/yr = 310 units/mo
(b) The actual (seasonalized) forecast is Ysz = (SI) Yc = (0.13)(310) = 40 units.

Exponential smoothening is a moving-average forecasting technique that weights past data in an

Exponential manner so that most recent data carry more weight in the moving average.
With simple Exponential smoothening, the forecast Ft is made up of the last period forecast
Ft–1 plus a portion, , of the difference between the last periods actual demand At–1 and last period
forecast Ft–1.
F t = Ft–1 + (At–1– F t–1).
I LLUSTRATION 6: A firm uses simple exponential smoothing with = 0.1 to forecast
demand. The forecast for the week of February 1 was 500 units, whereas actual demand
turned out to be 450 units.
(a) Forecast the demand for the week of February 8.
(b) Assume that the actual demand during the week of February 8 turned out to be 505
units. Forecast the demand for the week of February 15, Continue forecasting through
March 15, assuming that subsequent demands were actually 516, 488, 467, 554 and
510 units.
SOLUTION
(a) F t = Ft–1 + (At–1– Ft–1)
= 500 + 0.1(450 – 500) = 495 unit
(b) Arranging the procedure in tabular form, we have
 Table 5.4

Week Actual demand Old forecast Forecast error Correction New forecast
At – 1 Ft – 1 A t – 1 – Ft – 1 (At – 1 – Ft – 1) (Ft)Ft – 1 + (At – 1– Ft –1 )
Feb. 1 450 500 – 50 –5 495
8 505 495 10 1 496
15 516 496 20 2 498
22 488 498 – 10 –1 497
Mar. 1 467 497 – 30 –3 494
8 554 494 60 6 500
15 510 500 10 1 501

The smoothing constant, , is a number between 0 and 1 that enters multiplicatively into each
forecast but whose influence declines exponentially as the data become older. Typical values range
from 0.01 to 0.40. An Ion gives more weight to the past average and will effectively dampen high
random variation. High values are more responsive to changes in demand (e.g., from new-product
introductions, promotional campaigns). An of 1 would reflect total adjustment to recent demand,
and the forecast would be last period’s actual demand. A satisfactory can generally be determined
by trial-and-error modeling (on computer) to see which value minimizes forecast error.
Simple exponential smoothing yields only an average. It does not extrapolate for trend effects.
No value will fully compensate for a trend in the data. An value that yields an approximately
equivalent degree of smoothing as a moving average of n periods is:
2
=
n 1

Adjusted exponential smoothing models have all the features of simple exponential smoothing models,
plus they project into the future (for example, to time period t + 1) by adding a trend correction
increment, Tt, to the current period smoothed average, Fˆt .
Fˆt 1 = Fˆt Tt
Figure 5.1 depicts the components of a trend-adjusted forecast that utilizes a second smoothing
coefficient . The value determines the extent to which the trend adjustment relies on the latest
difference in forecast amounts ( Fˆ Fˆ ) versus the previous trend Tt–1 Thus:
t t 1

Fˆt = At 1 (1 ) Fˆt 1 Tt 1

Tt = ( Fˆt Fˆt 1 ) (1 )Tt 1

A low gives more smoothing of the trend and may be useful if the trend is not well-established.
A high will emphasize the latest trend and be more responsive to recent changes in trend. The
initial trend adjustment Tt–1 is sometimes assumed to be zero.
 Fig. 5.1 Components of trend adjusted forecast
Self-adaptive models: Self-adjusting computer models that change the values of the smoothing
coefficients s and s in an adaptive fashion have been developed; these models help to minimize
the amount of forecast error.

Regression and correlation techniques quantify the statistical association between two or more variables.
(a) 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.
(b) Simple correlation expresses the degree or closeness of the relationship between two variables
in terms of a correlation coefficient that provides an indirect measure of the variability of
points from the line of best fit. Neither regression nor correlation gives proof of a cause-effect
relationship.

The simple linear regression model takes the form Yc = a + bX, where Yc is the dependent variable
and X the independent variable. Values for the slope b and intercept are obtained by using the
normal equations written in the convenient form:

XY nXY
b = 2 (1)
X nX 2
a = Y bX (2)

In Equations. (1) and (2), X ( X ) / n and Y ( Y ) / n Y are the means of the independent
and dependent variables respectively, and n is the number of pairs of observations made.
I LLUSTRATION 7: The general manager of a building materials production plant feels
that the demand for plasterboard shipments may be related to the number of construction
permits issued in the county during the previous quarter. The manager has collected the
data shown in Table 5.5.
(a) Compute values for the slope b and intercept a.
(b) Determine a point estimate for plasterboard shipments when the number of construction
permits is 30.
Table 5.5

Construction permits (X) Plasterboard shipments (Y)

15 6
9 4
40 16
20 6
25 13
25 9
15 10
35 16

SOLUTION: (a)
Table 5.6

X Y XY X2 Y2
15 6 90 225 36
9 4 36 81 16
40 16 640 1,600 256
20 6 120 400 36
25 13 325 625 169
25 9 225 625 81
15 10 150 225 100
35 16 560 1,225 256
184 80 2,146 5,006 950

184
n = 8 pairs of observations X = X /n or X 23
8
80
Y = Y /n or Y 10
8
XY nXY 2146 8(23)(10)
b = 2 or b 0.395
X nX 2 5006 8(23)(23)
 a = Y bX or a 10 0.395(23) 0.91
(b) The regression equation is
Yc = 0.91 + 0.395 X (X = permits, Y = shipments)
Then, letting X = 30,
Yc = 0.91 + 0.395(30) = 12.76 ~ 13 shipments.
STANDARD DEVIATION OF REGRESSION
A regression line describes the relationship between a given value of the independent variable X and
y–x
the mean of the corresponding probability distribution of the dependent variable Y. We assume
the distribution of Y values is normal for any given X value. The point estimate, or forecast, is the
mean of that distribution for any given value of X.
The standard deviation of regression S y – x is a measure of the dispersion of data points around
the regression line. For simple regression, the computation of S y – x has n – 2 degrees of freedom.

Y2 a Y b XY
Sy x = .
n 2
I LLUSTRATION 8: Using the data from illustration 7, compute the Standard Deviation of
Regression.

SOLUTION

Y2 a Y b XY
Sy x =
n 2
950 (0.91)(80) (0.396)(2146)
Sy x = 2.2 shipments.
8 2
I LLUSTRATION 9: Using the data from illustrations 7 and 8,develop a 95 per cent prediction
interval estimate for the specific number of shipments to be made when 30 construction permits
were issued during the previous quarter.
Note: X = 23 for the n = 8 observations, and (X – X)2 = 774. Also, from Illustration 7,
Yc = 13 shipments, where X = 30; and from Illustration 8, Sy – x = 2.2 shipments.

SOLUTION
Prediction interval = Yc ± tSind (3)
where the t-value for n – 2 = 8 – 2 = 6 degrees of freedom = 2.45 and where

1 (X X )2
S ind = Sy x 1
n (X X )2

1 (30 23)2
S ind = 2.2 1 2.4 shipments
8 774
Production interval = 13 ± 2.45 (2.4) = 7.1 to 18.90 (use 7 to 19 shipments).
 For large samples (n 100), Equation 3 can be approximated by using the normal (Z) distribution
rather than the t, in the form of Yc ± ZSy–x (Note: For 95 per cent confidence, the Z value is the
same as t with df, which from Table 5.7 (given below) equals 1.96.) Also, the significance of the
regression line slope coefficient (b) can be tested using the expression:
b
tcalc = S
b
1
where Sb = S y x
( X X )2
If the value of tcalc > tdf from the t-table, the relationship between the X and Y variables is
statistically significant.

Table 5.7 -Distribution values (for 90 per cent and 95 per cent confidence)

df 5 6 7 8 9 10 12 15 20 30 00

(90%) 2.02 1.94 1.90 1.86 1.83 1.81 1.78 1.75 1.73 1.70 1.65
(95%) 2.57 2.45 2.37 2.31 2.26 2.23 2.17 2.13 2.08 2.04 1.96

The simple linear correlation coefficient r is a number between –1 and + 1 that tells how well a
linear equation describes the relationship between two variables. As illustrated in Fig. 5.2 r is designated
as positive if Y increases as X increases, and negative if Y decreases as X increases. An r of zero
indicates an absence of any relationship between the two variables.

Fig. 5.2 Interpretation of correlation coefficient

The deviation of all points (Y ) from the regression line (Yc ) consists of deviation accounted for
by the regression line (explained) and random deviation (unexplained). Fig. 5.3 illustrates this for
one point, Y. Squaring the deviation we have variation.
 Fig. 5.3 Deviation of dependent variable
Total variation = explained + unexplained
(Y Y )2 = (YC Y )2 (Y YC )2
The coefficient of determination r2 is the ratio of explained variation to total variation:

2 (YC Y )2
r =
(Y Y ) 2
The coefficient of correlation r is the square root of the coefficient of determination:
(YC Y )2
r =
(Y Y )2
When the sample size is sufficiently large (e.g., greater than 50), the value of r can be computed
more directly form:
n XY X. Y
r =
2
n X ( X ) n Y 2 ( Y )2
2

I LLUSTRATION 10: A study to determine the correlation between plasterboard shipments

X and construction permits Y revealed the following:
X = 184 Y = 80’ n=8
2 2
X = 5,006 Y = 950 XY = 2,146
Compute the correlation coefficient.

n XY X. Y
r =
n X2 ( X )2 n Y2 ( Y )2
8 (2146) (184) (80) 2448
r = 0.90.
[8(5006) (184)2 ][8(950) 802 ] 7430, 400

The significance of any value of r can be statistically tested under a hypothesis of no correlation.
To test, the computed value of r is compared with a tabled value of r for a given sample size and
significance level. If the computed value exceeds the tabled value, the correlation is significant.

A simple measure of forecast error is to compute the deviation of the actual from the forecast
values. Deviations will vary from plus to minus, but they should tend to average out near zero if the
forecast is on target.
Forecast error = actual demand – forecast demand.
The individual forecast errors are usually summarized in a statistic such as average error, mean
squared error, or mean absolute deviation (MAD).
Error
MAD =
n
The estimate of the MAD can be continually updated by using an exponential smoothing technique.
Thus the current MADt is:
MADt = (actual – forecast) + (1 – ) MADt–1
where is a smoothing constant. Higher values of will make the current MAD, more responsive
to current forecast errors.
When the average deviation (MAD) is divided into the cumulative deviation
[ (Actual forecast) ], the result is a tracking signal:
(Actual forecast)
Tracking signal =
MAD
Tracking signals are one way of monitoring how well a forecast is predicting actual values.
They express the cumulative deviation (also called the running sum of forecast error, RSFE) in
terms of the number of average deviations (MADs). Action limits for tracking signals commonly
range from three to eight. When the signal goes beyond this range, corrective action may be required.
ILLUSTRATION 11: A high-valued item has a tracking-signal-action limit of 4 and has
been forecast as shown in Table 5.8. Compute the tracking signal, and indicate whether some
corrective action is appropriate.
Table 5.8
Error (Error)2
Period Actual Forecast |Error|
(A – F) (A– F)
1 80 78 2 2 A
2 92 79 13 13 169
3 71 83 –12 12 144
4 83 79 4 4 16
5 90 80 10 10 100
6 102 83 19 19 361
Totals 36 60 794
 Error 60
MAD = 10
n 6
(Actual forecast) 36
Tracking signal = 3.6
MAD 10
Action limit of 4 is not exceeded. Therefore, no action is necessary.
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 .
(Actual Forecast )2
SF =
n 1
Control limits are then set, perhaps at two or three standard deviations away from the forecast
average X or the 2SF or 3SF limits are used as maximum acceptable limits for forecast error. Note
that the limits are based on individual forecast values, so you assume that the errors are normally
distributed around the forecast average.
I LLUSTRATION 12
(a) Compute the 2SF control limits for the data given in Illustration 11.
(b) Are all forecast errors within these limits?
(a) Control limits about the mean CL = X ± 2SF
Where
78 79 83 79 80 83
X = 80
6
(Actual Forecast)2 794
SF = 196 14
n 1 6 2
Therefore, CL = 80 ± 2(14) = 52 to 108 (rounded to integer values).
(b) All forecast errors (as calculated in Illustration 5.11) are within the ::t:: 28 error limit.
Note: Since n is less than 30, this distribution of forecast errors does not wholly satisfy the normality
assumption.
FORECAST APPLICATION
Forecasts should be sufficiently accurate to plan for future activities. Low-accuracy methods may
suffice; higher accuracy usually costs more to design and implement. Long-term forecasts—used
for location, capacity, and new-product decisions-require techniques with long-term horizons.
Short-term forecasts— such as those for production-and-inventory control, labour levels, and cost
controls–can rely more on recent history.
I LLUSTRATION 13: A food processing company uses a moving average to forecast next
month’s demand. Past actual demand (in units) is as shown in Table 5.9.