A robust estimator for location in Phase I based on an EWMA chart

In practice, a control chart for process monitoring (Phase II) is based on parameters estimated from data collected on the process characteristic under study (Phase I). The Phase I data could contain unacceptable data, which in turn could affect the monitoring. This article considers various estimation methods that are potentially relevant within the parameter estimation process. The quality of the Phase I study is evaluated in terms of the precision of the resulting estimates as well as the effectiveness of the exploratory data analysis, where effectiveness is measured by the proportion of observations that are correctly identified as unacceptable. Moreover, the impact of the Phase I estimation method on the performance of the EWMA control chart in Phase II is studied.