Determining out-of-control variable(s) in a multivariate quality control chart

Abstract

The Mason Young and Tracy (MYT) decomposition is considered to be the most  promising techniques for identifying variables that significantly cause abnormality in any monitoring process. This technique has been applied to two and three process variables. In this work, we extend the technique to four process variables which can be used for identifying out-of-control variables when signal occurs in a control chart. Twenty four (24) decompositions are presented of which one of the decompositions is used to identify out-of-control variables. Shift in the mean as well as in the variance-covariance structure is detected using the MYT decomposition model.

Key words: Hotelling’s T² control chart, MSQC library, Multivariate Statistical Process control, MYT decomposition.