An important by-product of a good forecasting procedure is to gain som
e understanding of how the series behaves. This paper looks primarily
at how well the Additive Holt-Winters method succeeds in this task. Ad
ditive Holt-Winters is a widely used technique which has had some succ
ess in forecasting competitions. The paper shows that the basic form o
f the method does not give good estimates for the level and seasonal f
eatures of a time series. The paper then looks at various ways of corr
ecting the estimates, looks at their performance and recommends a part
icular approach. (C) 1998 Elsevier Science B.V. All rights reserved.