On the Use of Principal Components for Assessing Multivariate Process Capability

This paper studies the effects of multivariate measurement errors on mul- tivariate capability indices computed using the principal components analysis. This study shows that measurement errors influence the results of a multivariate process capability analysis, resulting in either a decrease or an increase in the capability of the process. To avoid unreliable conclusions a method is proposed for overcom- ing the effects of measurement errors. Furthermore, a statistical test that allows one to determine whether measurement errors alter the process covariance structure is discussed