Biplots Variability based on the Procrustes Analysis

Multivariate analyses offer graphic representation for variables or objects in a reduced space(2-dimension) so that we can easily visualize their relationship. Since one can obtain the configurations of various form by algebraic algorithms, there exits a variability among configurations. The Procrustes analysis has been used mostly in multidimensional scaling. In this study, we applied the Procrustes analysis to biplots. As a result, if two plots showed differently, then the value of the Procrustes statistic was large. On the other hand, when two plots are similar, the Procrustes statistic was small. When the Procrustes statistic of two configurations is near to zero, we can consider two configurations are more similar. Thus, we can judge which configurations are very similar than other. we will demonstrate biplots variability for comparing configurations of biplots.