Abstract
Autocorrelated data present problems when traditional control charts are applied to the process or are modified to account for the autocorrelation. Forecast-based monitoring schemes have been suggested for use in these cases, but the resulting phenomenon of recovery of forecast errors can have a significant effect on the performance of the control chart chosen. To better enable practitioners to select a control chart for forecast based schemes, this article describes the dynamic mathematical behavior of one-step-ahead forecast errors for various AR (2). This behavior is used to explain the varying performance results of the Individuals, EWMA, and CES control charts applied to the one-step-ahead forecast errors. The article provides an introduction to autocorrelated processes, a description of various models and process shifts, a discussion of the impact of forecast recovery for the AR(2) model, realizations of the AR(2) process in regards to recovery rates and model parameters, a discussion of control chart performance, and conclusions.
Original language | American English |
---|---|
Journal | Palmetto Review |
Volume | 3 |
State | Published - 2000 |
DC Disciplines
- Applied Statistics
- Statistical Methodology
- Statistical Models