I have spectral data ranging from 200-2600nm.For building AAN models, I have used PCA as data reduction method to reduce the input variables. Based on eigenvalues>1, 14 PC was chosen. Then using graph builder I have plotted the 14 PC vs the Wavelength (200-2600nm) to see which PC's contribute more in the separation and which wavelengths are more dominant. Now I have been questioned how this can be seen from PCA? Enclosed is the plot.
So, I think you're doing a classic chemometrics experiement where you are want to use Raman or FTIR or some spectral method to characterize a sample with a known concentration of some analyte. Often the analyte is quantified by HPLC or some other more time consuming method. The goal is to make a model that will allow you to predict the quantity of analyte in the sample using the simple spectroscopic detection method.