The Detection of Shifts in Autocorrelated Processes with Moving Range and Exponentially-Weighted Moving Average Charts

The objective of this research is to select the appropriate control charts for detecting a shift in the autocorrelated observations. The autocorrelated processes were characterized using AR (1) and IMA (1, 1) for stationary and non-stationary processes respectively. A process model was simulated to achieve the response, the average run length (ARL). The empirical analysis was conducted to quantify the impacts of critical factors e.g., AR coefficient (φ), MA coefficient (θ), types of charts and shift sizes on the ARL. The results showed that the exponentially weighted moving average (EWMA) was the most appropriate control chart to monitor AR (1) and IMA (1, 1) processes because of its sensitivity. For non-stationary case, the ARL at positive θ was significantly higher than the one at negative θ when a shift size was small. If the performance of the statistical process control under stationary and nonstationary disturbances is correctly characterized, practitioners will have guidelines for achieving the highest possible performance potential when deploying SPC. The Detection of Shifts in Autocorrelated Processes with MR and EWMA Charts Karin KANDANANOND 16