Multivariate analysis of NMR spectra (0 replies)
 

Multivariate analysis of NMR spectra (0 replies)

Hi,
I am currently debating the benefit of different multivariate analysis methods for analysing NMR spectra and wondered if anyone had any ideas?

I have implemented principal components analysis (PCA) and canonical variates analysis (CVA). I am now looking into partial least squares discriminate analysis (PLS-DA).

My data contains a continuum of disease scores which I have categorised as Low Moderate and High. I would like to create a prediction model of disease state. The CVA analysis is supervised, so takes these categories into consideration however it sees them as independent groups. Where as the PLS-DA see Moderate as exactly half way between Low and High. Neither fits perfectly (as is expected with biological systems).

Does anyone have any comments on which is the most appropriate method, or could you point me in the direction of a paper or website that has easy to follow steps to learn PLS-DA? I have access to R, SAS and Genstat stats programmes.

Thanks for your help.
Pippa



More...