New Estimation Rules for Unknown Parameters on Holt-Winters Multiplicative Method

The Holt-Winters method is a well-known forecasting method used in time-series analysis to forecast future data when a trend and seasonal pattern is detected. There are two variations, i.e. the additive and the multiplicative method. Prior study by Vercher, et al. in [1] has shown that choosing the initial conditions is very important in exponential smoothing models, including the Holt-Winters method. Accurate estimates of initial conditions can result in better forecasting results. In this research, we propose new estimation rules for initial conditions for the Holt-Winters multiplicative method. The estimation rules were derived from the original initial conditions combined with the weighted moving average method. From the experimental results it was found that the new approach of the Holt-Winters multiplicative method can outperform the original Holt-Winters multiplicative method