Scribd
Upload
76 views

Forecasting Errors

The document discusses forecasting and methods for measuring forecasting errors. It defines forecasting as predicting future events and conditions. Organizations need forecasts to make intelligent decisions. The document then provides examples of areas where businesses require forecasts. It also discusses factors that influence forecasts and different methods for measuring forecasting errors, including mean absolute deviation, mean squared error, mean absolute percentage error, and mean percentage error.

Uploaded by

Sheikh Farhan Waheed
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
76 views

Forecasting Errors

The document discusses forecasting and methods for measuring forecasting errors. It defines forecasting as predicting future events and conditions. Organizations need forecasts to make intelligent decisions. The document then provides examples of areas where businesses require forecasts. It also discusses factors that influence forecasts and different methods for measuring forecasting errors, including mean absolute deviation, mean squared error, mean absolute percentage error, and mean percentage error.

Uploaded by

Sheikh Farhan Waheed
Copyright
© © All Rights Reserved
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 9

Forecasting

Prediction of future events and conditions are called forecasts, and


the act of making such predictions is called forecasting.

Any organization must be able to make forecasts in order to make


intelligent decisions. For instance, water quality, unemployment rate,
inflation rate, welfare payments, student enrolment, etc.

Business firms require forecasts of many events and conditions in all


phases of their operations.

Examples:

a) In marketing departments: Total demand for product must be


forecasted in order to plan total promotional effort.
b) In finance: Interest rates must be predicted to plan new capital
acquisitions.
c) In personnel management: Forecast the number of workers in
order to plan job requiting and training programs.
d) Process control: Forecast the behavior of the industrial process
in order to control the number of defective items.
e) Strategic management: Forecast of general economic conditions,
price and cost changes, technological change, and market growth in
order to plan long-term future of the company.

Information concerning events occurred in the past is very important


to learn in order to forecast about the future.
Errors in forecasting:
All forecasting situations involve some degree of uncertainty. The
presence of irregular component means that some error in forecasting
is expected. However, if the effect of irregular component is small,
then determination of appropriate trend, seasonal, or cyclical patterns
should allow us to forecast with accuracy. The large forecasting errors
may indicate that the irregular component is so large that no
forecasting technique will produce accurate forecasts, or may indicate
that the forecasting technique is not capable of accurately predicting
the trend, seasonal, or cyclical components.
Measuring forecast errors:
The forecast error for a particular forecast Y^ is e =Y −Y^ where Yt = actual
t t t t

value for time period t and Y^ is a predicted value for time period t.
t

If the forecasting technique is accurately forecasting the trend,


seasonal and cyclical components then the forecast errors should be
purely random and reflect only irregular components.
Absolute Deviation (AD)
In order to prevent positive and negative forecast errors from
cancelling each other out is to take absolute values of the forecasting
errors. The absolute values of forecasting errors are called absolute
deviations. That is given by: AD = |et | = |Y −Y^ |
t t

Mean Absolute Deviation (MAD)


The average of the absolute deviation of all the forecasts is known as
mean absolute deviation (MAD).

That is given by:


n
1
MAD= ∑ |Y −Y^ |
t t
ni=1
Actual Predicted Absolute
Value Value Error Deviation
yt y^t et =Y −Y^ t t |et| = |Y −Y^ |
t t

25 22 3 3
28 30 -2 2
29 30 -1 1
Totals 6

===n
n
1 6
MAD= ∑ |Y t −Y^t|= =2
n i=1 3

Squared Error (SE)


The other way to prevent positive and negative forecast errors from
cancelling each other out is to take square of the forecasting errors.
The squares of forecasting errors are called squared errors. That is
given by: SE = (et )2 = (Y −Y^ )2 t t

Mean Squared Error (MSE)


The average of the squared errors for all forecasts is called as Mean
Squared Error (MSE). Given the squared errors, we can define the mean
squared errors, such as:
n
1 2
MSE= ∑ ( Y t−Y^t )
n i=1

The forecaster would prefer smaller forecast errors to larger errors.


Example 2: Computation of Mean squared error
Actual Predicted
Value Value Error Squared Error
2
yt y^t et = Y −Y^ 2
(et) = (
t Yt− ^
Y ) t t

25 22 3 9
28 30 -2 4
29 30 -1 1
Totals 14

n
1 2 14
MSE= ∑ ( Y t−Y^t ) = =4.67
n i=1 3

Difference between MAD and MSE


The basic difference between MAD and MSE is that the MSE,
unlike MAD, penalizes a forecasting technique much more for
larger errors than for small errors.

The MSE and MAD can be used to:

1) Aid in the process of selecting a forecasting model.


2) Monitor a forecasting system in order to detect when something has
“gone wrong” with the system.
Example 3: Comparison of the errors produced by two different methods
Actual Predicted Absolute
Value Value Error Deviation Squared Error

2
yt y^t et = Y t −Y^t |et| = | (et)2 = (Y t −Y^t)
Y −Y^ |
t t

60 57 +3 3 9
64 61 +3 3 9
67 70 -3 3 9
Totals 9 27
n
1 9
MAD= ∑ |Y t −Y^t|= =3
n i=1 3
n
1 2 27
MSE= ∑ ( Y t−Y^t ) = =9
n i=1 3

Actual Predicted Absolute


Value Value Error Deviation Squared Error
2
yt y^t et = Y t −Y^t |et| = |Y t −Y^t| (et)2 = (Y −Y^ ) t t

60 59 +1 1 1
64 65 -1 1 1
67 73 -6 6 36
Totals 8 38

1
n
8 =
MAD= ∑ |Y t −Y^t|= =2.67
n i=1 3
n
1 2 38
MSE= ∑ ( Y t−Y^t ) = =12.67
n i=1 3
Example 3: Comparison of the errors produced by two different methods
Interpretation:
Forecasting method A has produced predictions yielding moderate
forecasting errors, while forecasting method B has produced predictions
yielding two small errors along with one large error. Forecasting method
A has larger MAD, while forecasting method B has the larger MSE. This
is so because in the calculation of the MSE, forecasting method B is
heavily penalized for its large error in forecasting the actual value 67.

Mean Percentage Error (MPE)


A way of measuring the forecasting error that allows comparison across
different time series with values of different magnitude is called mean
percentage error. This type of error can be calculated as:
n
1 (Y t −Y^t )
MPE= ∑
n i=1 Yt

Mean absolute percentage error (MAPE)


The average of the absolute percentage errors for all forecasts is known
as Mean absolute percentage error.
This can be given by:
n
1 |Y t−Y^t|
MAPE= ∑
n i=1 Y t
Example 4:
Actual Predicted Absolute
Value Value Error Percentage Error
yt y^t |et| = |Y −Y^ | |e | ×100
t t t

Yt
25 22 3 12.0
28 30 2 7.1
29 30 1 3.5
Totals 22.6
n
1 |Y t−Y^t|
MAPE= ∑
n i=1 Y t
¿ 7.533
Measuring Forecasting Error: The difference between observed values
and forecast value is called residua or forecasting error. That is
t e =Y −Y^ .
t t
Some of the formulae for evaluating forecasting errors are as follows:

Mean Absolute Deviation


n
1
MAD= ∑ |Y −Y^ |t t
n i=1

Mean Squared Error


n
1 2
MSE= ∑ ( Y −Y^ )
t t
n i=1

Mean Absolute Percentage Error


n
1 |Y t−Y^t|
MAPE= ∑
n i=1 Y t
Mean Percentage error
n
1 (Y t −Y^t )
MPE= ∑
n i=1 Yt

You might also like