Accuracy in Forecasting

FORECAST ACCURACY
• The difference between the forecast and actual demand.
• Although forecast error is inevitable, the objective of forecasting is
that it be as slight as possible.

• A large degree of error may indicate that either the forecasting

technique is the wrong one or it needs to be adjusted by changing its
parameters (for example, in the exponential smoothing forecast).
Continued…
• There are different measures of forecast error.
• We will discuss several of the more popular ones:
Mean absolute deviation (MAD),
Mean absolute percent deviation (MAPD),
Cumulative error,
and average error or bias.
MEAN ABSOLUTE DEVIATION
• The mean absolute deviation, or MAD, is one of the most popular
and simplest to use measures of forecast error.
• MAD is an average of the difference between the forecast and actual
demand, as computed by the following formula:
In discussed problem
• Forecasts were developed using exponential smoothing, ( α=0.30 and
α =0.50), adjusted exponential smoothing (α = 0.50, β= 0.30), and a
linear trend line, respectively, for the demand data for HiTek
Computer Services.
• The company wants to compare the accuracy of these different
forecasts using MAD.
Key Points
• The lower the value of MAD, relative to the magnitude of the data, the
more accurate the forecast.
• The MAD values for the remaining forecasts are as follows:
• Exponential smoothing (α = 0.50): MAD = 4.04
• Adjusted exponential smoothing (α = 0.50, β = 0.30): MAD = 3.81
• Linear trend line: MAD = 2.29
Outcome
• Since the linear trend line has the lowest MAD value of 2.29, it would
seem to be the most accurate.
• Although it does not appear to be significantly better than the adjusted
exponential smoothing forecast.
• Furthermore, we can deduce from these MAD values that increasing
from α= 0.30 to 0.50 enhanced the accuracy of the exponentially
smoothed forecast.
• The adjusted forecast is even more accurate.
Mean absolute percent deviation (MAPD)
• Since the linear trend line has the lowest MAD value of 2.29, it would
seem to be the most accurate.
• Although it does not appear to be significantly better than the adjusted
exponential smoothing forecast.
• Furthermore, we can deduce from these MAD values that increasing α
from 0.30 to 0.50 enhanced the accuracy of the exponentially
smoothed forecast.
• The adjusted forecast is even more accurate
MAPD: the absolute error as a percentage
of demand.
• The Mean absolute percent deviation (MAPD) measures the
absolute error as a percentage of demand rather than per period.
• As a result, it eliminates the problem of interpreting the measure of
accuracy relative to the magnitude of the demand and forecast values,
as MAD does.
• The mean absolute percent deviation is computed according to the
following formula:
 Using the data from the table in Previous
Example
• The exponential smoothing forecast ( α= 0.30) for HiTek Computer
Services:-
• MAPD = 53.39/557 = 0.096 = 9.6%
• A lower percent deviation implies a more accurate forecast.
• The MAPD values for our other three forecasts are:-
• Exponential smoothing (α = 0.50): MAPD = 7.9%
• Adjusted exponential smoothing (α = 0.50, β = 0.302) : MAPD = 7.5%
• Linear trend line: MAPD = 4.9%
CUMULATIVE ERROR
• Cumulative error: the sum of the forecast errors.
• 𝐸 = σ 𝑒𝑡
= 49.31
• This large positive error for cumulative error, plus the fact that the
individual errors for all but two of the periods in the table are positive,
indicates that this forecast is consistently below the actual demand.
Continued…
• A quick glance back at the plot of the exponential smoothing (α= 0.30)
forecast in Example visually verifies this result.
• The cumulative error for the other forecasts are:-
• Exponential smoothing (α = 0.50) : E = 33.21
• Adjusted exponential smoothing (α = 0.50, β = 0.30): E = 21.14
• We did not show the cumulative error for the linear trend line. E will
always be near zero for the linear trend line.
Average error: the per-period average of
cumulative error.
• A measure closely related to cumulative error is the average error, or
bias.
• It is computed by averaging the cumulative error over the number of
time periods:

E=49.32/11 = 4.48
 Comparison of Forecasts for HiTek Computer
Services

The average error is interpreted similarly to the cumulative error.

A positive value indicates low bias,
and a negative value indicates high bias. A value close to zero implies a lack of bias.
Key points
• Services example data, a larger value of α is preferable for the
exponential smoothing forecast.
• The adjusted forecast is more accurate than the exponential smoothing
forecasts,
• The linear trend is more accurate than all the others.
• Although these results are for specific examples, they indicate how the
different forecast measures for accuracy can be used to adjust a
forecasting method or select the best method.
 FORECAST CONTROL
• There are several ways to monitor forecast error over time to make sure
that the forecast is performing correctly—that is, the forecast is in control.
• Forecasts can go “out of control” and start providing inaccurate forecasts
for several reasons:-
A change in trend,
The unanticipated appearance of a cycle,
An irregular variation such as unseasonable weather, a promotional
campaign, new competition, or a political event that distracts consumers.
Tracking signal: monitors the forecast to see if it is
biased high or low.
• A tracking signal indicates if the forecast is consistently biased high or
low.
• It is computed by dividing the cumulative error by MAD, according to
the formula
Continued…
• The tracking signal is recomputed each period, with updated,
“running” values of cumulative error and MAD.
• Typically, forecast errors are normally distributed, which results in the
following relationship between MAD and the standard deviation of the
distribution of error, :

• 1 MAD = 0.8 σ
Key Points

• This enables us to establish statistical control limits for the tracking

signal that corresponds to the more familiar normal distribution.
• For example, statistical control limits of ±3 standard deviations,
corresponding to 99.7% of the errors, would translate to 3.75 MADs;
that is, 3 ÷ 0.8 = 3.75 MADs.
• Control limits of ±2 to ±5 MADs are used most frequently.
Interpretation
• The tracking signal values in the table above move outside 3 MAD
control limits (i.e., 3.00) in period 5 and continue increasing.
• This suggests that the forecast is not performing accurately or, more
precisely, is consistently biased low (i.e., actual demand consistently
exceeds the forecast).
• For the sake of comparison, the tracking signal for the linear trend line
forecast computed in Example is also plotted on this graph.
• Notice that it remains within the limits (touching the upper limit in
period 3), indicating a lack of consistent bias.
Another method for monitoring forecast error
is statistical control charts.
• For example, 3 control limits would reflect 99.7% of the forecast
errors (assuming they are normally distributed).
• The sample standard deviation, , is computed as

This formula without the square root is known as the mean squared error
(MSE), and it is sometimes used as a measure of forecast error.
Continued…
• Using the same example for the exponential smoothing forecast (
α=0.30) for HiTek Computer Services.
• we compute the standard deviation as
LINEAR REGRESSION
• Linear regression is a mathematical technique that relates one
variable, called an independent variable, to another, the dependent
variable, in the form of an equation for a straight line.
• A linear equation has the following general form:
• y = a + bx
• Where, y = the dependent variable
a = the intercept
x = the independent variable
Linear regression relates demand (dependent variable) to
an independent variable.
Example
• The State University athletic department wants to develop its budget
for the coming year using a forecast for football attendance. Football
attendance accounts for the largest portion of its revenues, and the
athletic director believes attendance is directly related to the number of
wins by the team. The business manager has accumulated total annual
average attendance figures for the past eight years.
Continued…
• Given the number of returning starters and the strength of the
schedule, the athletic director believes the team will win at least seven
games next year.
• Develop a simple regression equation for this data to forecast
attendance for this level of success.
Sol.
Continued…
• The data points with the regression line are shown in the following
figure.
• Observing the regression line relative to the data points, it would
appear that the data follow a distinct upward linear trend, which would
indicate that the forecast should be relatively accurate.
• In fact, the MAD value for this forecasting model is 1.41, which
suggests an accurate forecast.
CORRELATION
• Correlation: a measure of the strength of the relationship between
independent and dependent variables.
• The value of r varies between -1.00 and +1.00, with a value of 1.00
indicating a strong linear relationship between the variables.
This value for the correlation coefficient is very close to
1.00, indicating a strong linear relationship
between the number of wins and home attendance.
Coefficient of determination: the percentage of the variation in
the dependent variable that results from the independent
variable.
This value for the coefficient of determination means that
89.7% of the amount of variation in attendance
can be attributed to the number of wins by the team (with
the remaining 10.3% due to
other unexplained factors, such as weather, a good or
poor start, or publicity). A value of 1.00 (or
100%) would indicate that attendance depends totally on
wins.
However, since 10.3% of the variation
is a result of other factors, some amount of forecast error
can be expected.
Problem
• Healthy Hamburgers has a chain of 12 stores in northern Illinois. Sales
figures and profits for the stores are given in the following table.
Obtain a regression line for the data, and predict profit for a store
assuming sales of $10 million.
 MULTIPLE REGRESSION WITH EXCEL
a relationship of demand to two or more independent variables.

• Another causal method of forecasting is multiple regression, a more powerful extension of

linear regression.
• Linear regression relates demand to one other independent variable, whereas multiple
regression reflects the relationship between a dependent variable and two or more independent
variables.
• A multiple regression model has the following general form:
Example Developing a Multiple Regression
Forecast with Excel
• To demonstrate the capability to solve multiple regression
problems with Excel spreadsheets, we will expand our State
University athletic department previous Example for
forecasting attendance at football games that we used to
demonstrate linear regression. Instead of attempting to predict
attendance based on only one variable, wins, we will include
a second variable for advertising and promotional
expenditures as follows:
We will use the “Data Analysis”
option (add-in) from the Tools menu
at the top of the
spreadsheet that we used in the
previous section to develop our linear
regression equation,
Sol. From excel
• However, as we have already noted, the number of wins would appear
to
• probably account for a larger part of the variation in attendance.
• = 46,229.35
• y = 19,094.42 + 3560.99x1 + 0.0368x2