Linear discriminants and Principal Component Analysis

Seldom is a single linear discriminant adequate for a real problem. Sometimes, we can use multiple linear discriminants where a single one would fail. The most popular way of doing this is Principal Component Analysis.

For a tutorial on those matters see

http://www.isip.msstate.edu/publications/reports/isip_internal/1998/linear_discrim_analysis/lda_theory.pdf

For something provably far superior to Principal Component Analysis see

”Independent task Fourier filters,” Opt. Engr. 40, 2414-2418 (2001), H. John Caulfield