I have a dataset of 240 NIR spectra, that is 240 observations of 300 highly correlated variables. I have read that the successive projections algorithm (SPA) is a good way to select informative variables in order to perform further analysis such as LDA (Linear Discriminant Analysis) but I am confused as to how to use it.
I have read this R vignette: https://cran.r-project.org/web/packages/lintools/vignettes/project.html but I do not understand what my restrictions would be (A and b in this case) or how this function would get me a variable selection.
I have also encountered this solution in matlab: http://www.ele.ita.br/%7Ekawakami/spa/SPA_GUI_Manual_v3p3.pdf but I am corious as to how it works and how to do it in R.
My reference papers are these two:
de Carvalho, L.M., et al. (2020), Occurrence of wooden breast and white striping in Brazilian slaughtering plants and use of near-infrared spectroscopy and multivariate analysis to identify affected chicken breasts. Journal of Food Science, 85: 3102-3112. https://doi.org/10.1111/1750-3841.15465
Ugulino Araújo, M.C., et al. (2001). The successive projections algorithm for variable selection in spectroscopic multicomponent analysis, Chemometrics and Intelligent Laboratory Systems, 57(2): 65-73, https://doi.org/10.1016/S0169-7439(01)00119-8.
