478 Statistical Methods
From the rescaled plot, you would conclude that exponential smoothing
has done a good job of modeling the sales data and that the resulting fore-
casts appear reasonable.
The fi nal part of the exponential smoothing output is the fi nal estimate of
the seasonal indexes, shown in Figure 11-34.
Figure 11-34
Seasonal
indices
The values for the seasonal indexes are very similar to those you calcu-
lated using the seasonal adjustment command. The difference is due to the
fact that these seasonal indexes are calculated using a smoothed average,
whereas the earlier indexes were calculated using an unweighted average.
You’re fi nished with the workbook. You can close it now, saving your
changes.
Figure 11-33
Plot of
forecasted
and observed
sales for the
current and
upcoming year