Control Procedures for Residuals Associated With Principal Component Analysis

摘要: This paper is concerned with the treatment of residuals associated principal component analysis. These are difference between original observations and predictions them using less than a full set components. Specifically, procedures proposed for testing single observation vector an overall test group observations. In this development, it assumed that underlying covariance matrix known; reasonable many quality control applications where may be quite useful in detecting outliers data. A numerical example included.