Speaker
Description
Hotelling’s $T^2$ control chart is probably the most widely used tool in detecting outliers in a multivariate normal distribution setting. Within its classical scheme, the unknown process parameters (i.e., mean vector and variance-covariance matrix) are estimated via a phase I (calibration) stage, before online testing can be initiated in phase II. In this work we develop the self-starting analogue of Hotelling’s $T^2$, within the Bayesian arena, allowing online inference from the early start of the process. Both mean and variance-covariance matrix will be assumed unknown, and a conjugate (power) prior will be adopted, guaranteeing a closed form mechanism. Theoretical properties, including power calculations of the proposed scheme, along with root-cause related post-alarm inference methods are studied. The performance is examined via a simulation study, while some real multivariate data illustrate its use in practice.
| Classification | Mainly methodology |
|---|---|
| Keywords | Bayesian statistical process control and monitoring, multivariate power prior, post alarm inference |