Multivariate GR&R through factor analysis

Abstract

Several measurement tasks present multivariate nature. In the cases with quality characteristics highly correlated within groups, but with a relatively small correlation between groups, the available multivariate GR&R methods are not suitable to provide a correct interpretation of the results. The present work presents a new multivariate GR&R approach through factor analysis. Factor analysis is a multivariate statistical method which focuses on the explanation of the covariance structure of the data. Through orthogonal rotation of the factors a suitable structure can be achieved with loadings easy to relate the variables to the factors. The proposed multivariate GR&R method through factor analysis is described and applied in the quality evaluation of holes obtained through helical milling process of AISI H13 hardened steel. The method succeeded in achieving a simple structure, with one factor related to the roughness outcomes and other related to the roundness ones, simplifying the gage capability evaluation.