Just What Does an EWMA Do (Part 2)? Mini Paper Taken from the Fall 1995 Newsletter
Abstract: The Exponentially Weighted Moving Average (EWMA) chart is an alternative to Shewhart control charts when the chart should permit a varying mean and correlated errors. When the system should have a constant mean, the EWMA smooths the historical time trace enabling the investigator to observe alternative paths for the mean. The search for assignable causes is often enhanced by review of the time trace of an EWMA chart. An EWMA chart can play an important role in the active control of industrial processes. Whenever a data trace Yt demonstrates autocorrelation the EWMA chart can provide an opportunity to reduce variability about target. The EWMA chart is able to forecast by viewing recent time history through a narrow window and by down weighting older observations. It is thus an appropriate statistic to employ when a process is non-stationary. It can also be used as a useful approximate forecast for many postulated dynamic models for the mean. Shewhart and EWMA charts can be complementary.
Keywords: Exponentially Weighted Moving Averages (EWMA) - Control charts - Shewhart chart - Average run length (ARL) - Statistical process control - Automatic process control - Autocorrelation - Variogram