BIPLOT ANALYSIS IN CLASSIFICATION

Abstract

The Biplot technique is presented as a graphical multivariate tool useful in classification of taxonomic units. The method is particularly useful in the display and analysis of large data matrices where visual examination is impractical. The Biplot obtains a two dimensional approximation to a matrix and allows inspection of variances, approximate correlations, relationships, patterns and clusters existing in the data. The basic theory related to Principal Component Analysis and the stucture of a data matrix is presented. A program was written in the matrix languaje of the Statistical Analysis System (IML/SAS) and the method of Biplot analysis is illustrated with an example.