I have a transmission FTIR data set that I will vector normalize before doing statistical analysis on it. For the statistical analysis, I am keeping only these regions 1612-925, 1797-1681 and 3060-2803 cm-1. The regions that I am discarding are either noisy due to water or they don’t contain any chemical information related to my samples.
My question is: will it be acceptable to calculate the spectrum norm and vector normalize using only the spectral variables that will be included in the statistical analysis, ie 1612-925, 1797-1681 and 3060-2803 cm-1? Does the discontinuity of the normalization range represent an issue for the normalization pre-treatment in this case?
Thank you