In today’s world, the amount of available data is steadily increasing, and it is often of interest to detect changes in the data. Statistical process monitoring (SPM) provides tools to monitor data streams and to signal changes in the data. One of these tools is the control chart. The topic of this dissertation is a special control chart: the exponentially weighted moving average (EWMA) chart.
A control chart plots the data together with two control limits. A control chart signals a (possible) change when the plotted data exceeds the control limits. A control chart performs well if it signals changes in the data quickly, without triggering frequent false alarms.
Before a control chart can be set up, estimates of the process parameters are needed. To this end an initial data set is collected. In practice this data set often contains outliers, recording errors, and other data quality issues. These so-called ‘contaminations’ are problematic as they influence the parameter estimates. We investigate robust estimation methods to ensure accurate estimation of the process parameters. We propose a new estimation method based on screening and show that this new method outperforms existing estimation methods, when the type of contaminations is unknown.
In the second phase of this dissertation we study the effect of estimation on the performance of the EWMA chart and give recommendations regarding its design. We show that traditionally designed charts have very variable performance. We study an alternative design procedure based conditional performance which provides control over the variability in performance