Principal Components Factor Analysis In Mineral Processing

Principal Components Analysis (PCA) is a versatile, multivariate statistical method that the Bureau of Mines is applying to mineral processing data to achieve data simplification and pattern recognition. The principal components are obtained from an eigenanalysis of the data matrix. To achieve data simplification PCA replaces a larger number of measured variables with a smaller set of principal components and at the same time removes a significant amount of experimental error. Data simplification and error reduction make PCA useful in process control. Plots of the individual principal components form the basis of a pattern recognition technique that allows detection of significant variations in the structure of the data. These applications of PCA are illustrated with appropriate mineral processing data.