Abstract

The principal components interpretation is generally based in the respective magnitudes of the values assigned to the variables. The simplification of this process has been for a long time a big preoccupation for the experts. For example, Jeffers [2] proposes an empirical rule of decision that consists on fix to zero the lowest values. Other popular methods are based in the rotation of the axis obtained for the PCA (Harman [3]). Most recently, Vines [4] propose a proceeding more sophisticated that reach a simplification of the principal components searching those can be represented by integer numbers. Vigneau y Qannari [5] discusses a strategy of simplification based in the analysis of a variable cluster.