Brown's
modified approach to exponential smoothing is useful when trend is evident in
the time series. The method introduces
logic similar to that found in moving averages. The Brown technique simply extends the single exponential technique
to a double exponential technique. Double
exponential smoothing is defined as a single exponential smoothing
applied to an already exponentially smoothed time series. Stated differently, it is the single
exponential smoothing of the single exponential values.
Similar
to the Unadjusted technique discussed above, the Brown's
technique requires the following three data items.
1 Alpha; where alpha is between 0 and 1 is associated with exponential adjustments to the permanent component (new value) of the next period forecast.
2 Time Period: 1, 4, or 12.
3 Initial Base Value for Time Series - optional.