Improvement to the Trend-Adjusted Exponential Smoothing calculation (17010)

In previous releases, if you applied Trend-Adjusted Exponential Smoothing (TAES) to a forecast for the upcoming week, the software analyzed the last reported 15 data points for Actual KBIs—that is, the prior 15 weeks of data—and calculated a trend factor that was applied to the upcoming week. When forecasting more than one week in the future, the software analyzed the data set (the prior 15 weeks of Actual KBI data) and assumed that both the data set and the trend factor for the first upcoming week would be identical for all future weeks in the desired forecast range.

In this release, the TAES modeling algorithm has been improved to analyze the prior 15 weeks of Actual KBI data, calculate a trend factor for the first week in the forecast range, and then extrapolate the calculation across the forecast range.

EXAMPLE

The TAES-calculated trend factor of 20 is applied to an 8-week forecast range.

Before the improvement:

TAES weighted average	TAES trend factor	Weeks in desired forecast range
380	20	1	2	3	4	5	6	7	8
		380+20	380+20	380+20	380+20	380+20	380+20	380+20	380+20
		400	400	400	400	400	400	400	400

With the improved calculation:

TAES weighted average	TAES trend factor	Weeks in desired forecast range
380	20	1	2	3	4	5	6	7	8
		380+20	400+20	420+20	440+20	460+20	480+20	500+20	520+20
		400	420	440	460	480	500	520	540

For more information on TAES and its application to statistical KBIs, see the following:

Updated versions: Version 9.15 and later.

Browser not supported