Linear Discriminant Analysis (LDA)
LDA, originally proposed by Fisher in 1936, is the first multivariate classification method in the literature. Decision surfaces separating the categories in the multivariate space are defined as linear combinations of the original variables (hyperplanes).
Details on LDA can be found in many textbooks and tutorials on supervised pattern recognition, so that, below, just a few of the possible references will be suggested.
The original paper:
R.A. Fisher, The Use of Multiple Measurements in Taxonomic Problems, Ann. Eugen. 7 (1936) 179-188.
A couple of tutorials on classification methods from our group:
M. Cocchi, A. Biancolillo, F. Marini, Chemometric Methods for Classification and Feature Selection. In: J. Jaumot, C. Bedia, R. Tauler (Eds.), Data Analysis for Omic Sciences: Methods and Applications, Comprehensive Analytical Chemistry, vol. 82, Elsevier, Amsterdam, The Netherlands, 2018, 265-299.
A. Biancolillo, F. Marini, Chemometric Classification Methods in Omic Data Analysis. In: C. Leon, A. Cifuentes (Eds.), Omics Data Treatment, System Biology and Foodomics, Elsevier, Amsterdam, The Netherlands, 2020, in press.
Download the matlab functions for calculating and validating LDA models, including the extraction of canonical variates.