*
* Example 9.3.1 from page 327
*
open data deaths.tsm
calendar(m) 1973
data(format=free,org=columns) 1973:1 1978:12 deaths
*
* The ESMOOTH instruction allows you to select from among three models for the
* trend (none, linear and exponential) and three for the seasonal (none, additive
* and multiplicative). The automatic selection procedure optimizes the
* coefficients in each and picks the model with the smallest BIC (here labeled as
* SBC - Schwarz Bayesian Criterion). The CONSTRAIN option forces the coefficients
* into the range [0,1]. (Values outside that can still create stable models).
* Note that RATS uses different Greek letters for the trend and seasonal
* smoothing parameters: gamma for trend (rather than beta) and delta for seasonal
* (rather than gamma).
*
esmooth(trend=select,seasonal=select,constrain,initial=start,forecasts=forecast,steps=24) deaths
*
* The Holt-Winters seasonal model used in example 9.3.1 is the combination of
* TREND=LINEAR and SEASONAL=ADDITIVE.
*
esmooth(trend=linear,seasonal=additive,constrain,initial=start,forecasts=forecast,steps=24) deaths
*
* Graph the actual data and the forecasts
*
graph 2
# deaths
# forecast
*
* Pull in the extra data for comparison
*
data(unit=input) 1979:1 1979:6 adeaths
 7798 7406 8363 8460 9217 9316
@uforeerrors adeaths forecast 1979:1 1979:6
print(picture="*.") 1979:1 1979:6 adeaths forecast