*
* Example TABLE6-7.DAT from pp 222-224
*
open data table6-7.dat
data(format=free,org=columns) 1 30 demand
*
esmooth(trend=linear,alpha=0.2,gamma=0.6,initial=start,smoothed=smoothed,fitted=fitted) demand
set error = demand-fitted
print(picture="*.###") / demand smoothed fitted error
*
graph(footer="Figure 6-5 Holt's two-parameter exponential smoothing",key=below) 3
# demand
# smoothed
# fitted
*
* You can optimize the parameters on the Holt model by using the option ESTIMATE.
* We also will use the option CONSTRAIN, which will force the smoothing
* parameters to lie in the unit interval. The ES recursions are stable outside
* that range, so RATS (by default) allows larger values.
*
esmooth(trend=linear,estimate,constrain,initial=start) demand