What we didn't cover
In an effort to spend more time laying a foundation and facilitating a deeper understanding of one of the most popular intermediate methods of forecasting (and, even as it is, I couldn't go into nearly as much detail as I would have liked to for want of space), we necessarily had to miss out on a few topics that would have been nice and helpful to cover. Particularly, the primary topic that comes to mind is ARIMA, or autoregressive integrated moving average, models.
ARIMA models, like exponential smoothing methods as of the late last and early this century, are often expressed as state space models. In addition, many of the exponential smoothing methods we used in this chapter can be translated to equivalent ARIMA models (this is, in fact, how many software programs provided prediction intervals for exponential smoothing forecasts before the state-space/ETS framework opened up the door to do it directly). Due to this translatability for certain popular models, I felt...
