For metabolomic data analysis, different choices for scaling are available such as Pareto, centering, unit variance etc. which of these scaling methods is recommended for FTIR data?
Thank you.
Most research groups use univariate scaling, predominantly because of the large dynamic range of the spectral variables or data points. However, if you are using a selected region of the spectrum where the variables (data points) are all of the same magnitude then mean centring without scaling or pareto scaling would be appropriate.
Duarte et al J. Agric. Food Chem., 2004, 52 (5), pp 1031–1038
DOI: 10.1021/jf030659z
This article is particularly good wrt preprocessing and scaling of FTIR data
Wehrli et al 2014, J Chemometrics DOI: 10.1002/cem.2588
Dear Dr. Elaine Holmes,
I appreciate your response. I found it very informative and I will refer to the papers you recommended for me
Thank you very much
- Badges
- Science topic