Automating Vibrational Spectroscopy Data Preprocessing and Multivariate Analysis with MATLAB

Abstract

The tutorial is structured according to the steps of preprocessing and multivariate analysis. We will, therefore, start with a discussion on how to import a spectrum into MATLAB, followed by how to derivatize this spectrum, select a specific spectral range from the derivative spectrum, and normalize the derivative spectrum of the selected range in Section 4. Section 5 will demonstrate how to automate the preprocessing by scripting to import multiple spectra and applying preprocessing commands to them. Section 6 will extend this concept to import multiple spectra from multiple subfolders and preprocess them all. Section 7 will provide details on applying PCA, PC-LDA, and leave-one-out cross validation (LOOCV) on the preprocessed spectra with a note on support vector machines (SVMs). Section 8 teaches scripting of test prediction code. Sections 9 and 10 will discuss automating PCA plotting and script fine-tuning to build in options to turn features on and off. Common errors encountered while executing the code and an example of how to adapt the code to automate other processes such as mean and standard deviation calculations are included at the end.