Figure 1. Two dimensional data set with the using of different algorithms.

Black line with a square represents the direction of the LDA which coincides exactly with the direction of LDA-PLS when the latent variable is equal to 2 (the proof can be seen from Section 3.2). Red dotted lines of and are the boundary lines which can be correctly separate the two kinds of samples, and all lines between these boundaries can also do. Boundary classification and correspond to the direction of ex-LDA-PLS () and ex-LDA-PLS ().