ESTIMATING THE PERIOD OF A STEP CHANGE IN SINGLEOBSERVATION DATA

摘要

Quality control charts are one of the standard statistical process control (SPC) techniques, which are effective in detecting special causes. When chart signals that a process gets out-of-control, a search begins to identify and remove the source(s) responsible for the process change. Knowing when the process first changed can help process engineers to identify and eliminate or optimize the variables responsible for the process change, which lead to poor quality. The exact time when process has changed is called change point. This paper provides an estimator for identifying the actual time of change in normal process mean and it is used with Tukey's control chart and Shewhart individual control chart. In many situations, it is necessary to monitor individual observations. The simulation result shows that the proposed estimator performs much better to identifying the actual time of the change in the process.