More than 2 years ago I published on this website a working paper entitled “Multiplicative State-Space Models for Intermittent Time Series“, written by John Boylan and I. This was an early version of the paper, which we submitted to International Journal of Forecasting on 31st January 2017. More than two years later (on 11th July 2019, when I was attending useR! conference, in fact 10 minutes before my presentation), after several serious revisions of the paper and a couple of radical updates of the code of `es()`

and `oes()`

functions from smooth package for R, the paper was rejected by an Associate Editor.

After that John Boylan and I have agreed to rewrite it one more time and submit to a different journal. And while we do that, I have decided that the last version of the paper should become freely available, because the models described there are implemented in `es()`

and `oes()`

functions, which I do not intent to rewrite anymore. The new paper will be supported by new functions, but it makes sense to have something, to refer to for the explanation of how `es()`

and `oes()`

work. So, here it is.