On the Orthogonal Decomposition of Experimental Data using Eigenvector Methods

In many situations involving analysis of experimental data, there exists a degree of redundancy. It is advantageous to identify and reduce the redundant elements of the data in order to reduce the dimensionality of the data, thus enabling better modelling or representation. The subsequent modelling may be performed using linear or nonlinear techniques, whose e ciency may be aided by the preprocessing described here.