Principal component analysis is applied to analyse the Raman maps collected on carbon nanotubes at different degrees of oxidation and functionalization with dye labeling molecules. The results are used to demonstrate that the technique is extremely effective in clustering data and comparing preparation protocols, so that it enables drawing of a fast and reliable classification of the molecule propensity to interact with pristine and oxidized carbon nanotubes. The spectral findings are supported and elucidated by several experimental techniques, thermogravimetry and steady-state and time-resolved fluorescence measurements, and by computational modeling, showing that the proposed methodology could represent a powerful and routine test for the rational design of functional nanostructures.