Abstract
We often analyze data in manufactures by the canonical correlation analysis to comprehend the dependence of effects upon causes, or the dependence of physical properties upon sensitive properties. In the canonical correlation analysis, we use linear combinations of effects. However, we are frequently interested in patterns of effects, and in the dependence of the patterns upon causes. In this paper, we propose a clustering method with covariates which can analyze the relationship between clusters of patterns of effects and the corresponding clusters of causes, and investigate characteristics of the proposed method.
