ABSTRACT
ABSTRACT The frequentist Shewhart charts have proved valuable for the first stage of quality improvement in many manufacturing settings. However, their statistical foundation is on a model with exactly known process parameters and independent identically distributed process readings. One or more aspects of this foundation are often lacking in real problems. A Bayesian framework allowing an escape from the independence and the known-parameter assumptions provides a conceptually sounder and more effective approach for process control when one moves away from this first idealization of a process.
