Forecast Reliability
Forecast reliability is used to measure the closeness of the forecast in reflecting the reality.
It is known that a forecast may not always be accurate. However, the primary objective of
forecasting is to make it generally reliable. One way of testing the reliability of the forecast is by
computing the mean absolute deviation (MAD).
Mean Absolute Deviation is the measure of the difference between a forecast and the
actual data/value.
The formula for the Mean Absolute Deviation is:
𝛴(𝑎𝑐𝑡𝑢𝑎𝑙 𝑑𝑎𝑡𝑎 − 𝑓𝑜𝑟𝑒𝑐𝑎𝑠𝑡)
MAD =
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑒𝑟𝑖𝑜𝑑𝑠
Table 5 and 6 shows the computation of the Mean Absolute Deviation of Simple
Exponential Smoothing forecast and Adjusted Exponential Smoothing forecast. Before you get
the MAD, you must compute the difference between the actual data (demand) and the forecast
per period or the Absolute error (|Error|). Absolute error means that regardless if the original value
is positive or negative, it should be written as positive.
The forecast may be the simple moving average, weighted moving average, simple
exponential smoothing or the adjusted exponential smoothing. It depends on what forecast you
are trying to check. In this case, we will check the reliability of the simple moving average forecast
and the adjusted exponential smoothing forecast. After computing for the errors, you must
compute for the sum of all the errors and divide it to the number of periods.
Example: |Error| = Y (demand) – F (forecast)
February =70 – 60 = 10
March = 50 – 61 = -11 (It should be written as positive 11 as we are getting the
absolute value.)
See below table to see computation for mean absolute deviation.
Forecast reliability is used to measure the closeness of the forecast in reflecting the reality.
It is known that a forecast may not always be accurate. However, the primary objective of
forecasting is to make it generally reliable. One way of testing the reliability of the forecast is by
computing the mean absolute deviation (MAD).
Mean Absolute Deviation is the measure of the difference between a forecast and the
actual data/value.
The formula for the Mean Absolute Deviation is:
𝛴(𝑎𝑐𝑡𝑢𝑎𝑙 𝑑𝑎𝑡𝑎 − 𝑓𝑜𝑟𝑒𝑐𝑎𝑠𝑡)
MAD =
𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑝𝑒𝑟𝑖𝑜𝑑𝑠
Table 5 and 6 shows the computation of the Mean Absolute Deviation of Simple
Exponential Smoothing forecast and Adjusted Exponential Smoothing forecast. Before you get
the MAD, you must compute the difference between the actual data (demand) and the forecast
per period or the Absolute error (|Error|). Absolute error means that regardless if the original value
is positive or negative, it should be written as positive.
The forecast may be the simple moving average, weighted moving average, simple
exponential smoothing or the adjusted exponential smoothing. It depends on what forecast you
are trying to check. In this case, we will check the reliability of the simple moving average forecast
and the adjusted exponential smoothing forecast. After computing for the errors, you must
compute for the sum of all the errors and divide it to the number of periods.
Example: |Error| = Y (demand) – F (forecast)
February =70 – 60 = 10
March = 50 – 61 = -11 (It should be written as positive 11 as we are getting the
absolute value.)
See below table to see computation for mean absolute deviation.
