Exponential Smoothing
AdvancedExponential smoothing is a forecasting method that predicts future values by taking a weighted average of past observations, giving the most weight to recent data and progressively less weight to older data.
Simple moving averages give equal weight to each data point in a window. Exponential smoothing instead gives recent observations more influence, so forecasts react faster to changes while still smoothing out random noise.
In simple exponential smoothing, each new forecast equals the previous forecast plus a fraction, called the smoothing factor, of the latest forecast error. For example, with a smoothing factor of 0.3, if last period's forecast was 100 and the actual value was 110, the next forecast is 100 plus 0.3 multiplied by 10, or 103.
A higher smoothing factor makes forecasts respond more quickly to recent changes but also to random noise. A lower factor produces smoother, steadier forecasts that adjust more slowly.
Extensions of the method handle trends and seasonal patterns, such as the Holt-Winters method. Exponential smoothing is widely used in sales forecasting, inventory planning and demand forecasting because it is simple, needs little data and works well for many business time series.
For worked examples and full lessons, visit notesbysara.org.