In time series forecasting, a moving average is a smoothing method in which the forecast for time t is the average value for the w periods ending with time t-1. The analyst sets w, the “window width” often with a view to encompassing all relevant seasons so that seasonality is suppressed.
