A Bayesian Approach to Multivariate Measurement System Assessment

Abstract

This article considers system assessment for multivariate measurements and presents a Bayesian approach to analyzing gauge R & R study data. The evaluation of variances for univariate measurement becomes the evaluation of covariance matrices for multivariate measurements. The Bayesian approach ensures positive definite estimates of the covariance matrices and easily provides their uncertainty. Moreover, various measurement-system assessment criteria are easily evaluated. The approach is illustrated with data from a real gauge R & R study as well as simulated data.

• Gauge R & R Study
• Markov-Chain Monte Carlo
• Multivariate Normal Distribution
• Random Effects
• Repeatability
• Reproducibility
• Scaled Inverse Wishart Prior Distribution
• Uncertainty
• Variance Components

Additional information

Notes on contributors

Michael S. Hamada

Dr. Hamada is a Scientist at Los Alamos National Laboratory. He is a Senior Member of ASQ. His email address is [email protected].