Starting Slow with Simple Exponential Smoothing
Exponential smoothing techniques base a future forecast off of past data where the most recent observations are weighted more than older observations. This weighting is done through smoothing constants. The first exponential smoothing method you're going to tackle is called simple exponential smoothing (SES), and it uses only one smoothing constant, as you'll see.
Simple exponential smoothing assumes that your time series data is made up of two components: a level (or mean) and some error around that level. There's no trend, no seasonality, just a level around which the demand hovers with little error jitters here and there. By preferring recent observations, SES can account for shifts in this level. In formula-speak then, you have:
- Demand at time t = level + random error around the level at time t
And the most current estimate of the level serves as a forecast for future time periods. If you're at month 36, what&apos...