*
* Example SERIESF.DAT from pp 227-229
*
open data seriesf.dat
data(format=free,org=columns) 1 48 demand
*
* There's quite a bit going on with this instruction. Because there is a seasonal,
* something has to tell ESMOOTH how long the seasonal is. If this were a true
* data set, you would probably have set a CALENDAR which would set the seasonality
* of the data, so you wouldn't need a SPAN option. The * 36 parameters have it
* smooth through period 36; steps=12 requests 12 steps of forecast beyond that.
* Note that the forecast option saves the out-of-sample forecasts, which will be
* the y-hat column in table 6-12.
*
esmooth(trend=linear,seasonal=multiplicative,span=12,alpha=0.3,gamma=0.05,delta=0.75,initial=start,smoothed=smoothed,$
 fitted=fitted,resids=errors,factors=seasonal,forecast=forecast,steps=12) demand * 36
*
print(picture="*.###") 1 36 demand smoothed seasonal fitted errors
print(picture="*.###") 37 48 demand forecast
@uforeerrors demand forecast 37 48
*
graph(footer="Figure 6-6. Winter's three parameter exponential smoothing",key=below) 3
# demand
# fitted
# forecast