2.2.4 Smooth

The choice of smoothing method used depends on the characteristics of the time series. If the mean of the series is only slowly changing and there is no seasonal component, then single exponential smoothing may be suitable. If there is a trend in the time series, which itself may be slowly changing, then double exponential smoothing may be suitable. If there is a seasonal component to the time series, e.g., monthly or yearly data, then Winter's methods may be suitable.

No trend, no seasonality Moving Average, Single Exponential Smoothing
With trend, no seasonality Double Exponential Smoothing
With seasonality(with or without trend) Winter's Method


Topics covered in this section: