ABSTRACT

Multivariate data analysis methods comprise a group of statistical and mathematical techniques that analyze multiple variables simultaneously. With the fusion of appropriate multivariate methods, hyperspectral imaging answers the questions about the sample such as what chemical species are in the sample, how much of each is present, and most importantly, where they are located. Multivariate methods decompose complex multivariable data into simple and easily interpretable structures to better understand the chemical and biological information of the tested samples. Once the spectral data is available for multivariate modeling, it is necessary to mitigate the noise from the data by applying pre-processing to obtain a good and robust prediction model. Multivariate techniques typically applied in spectral data can be divided into multivariate classification for qualitative analysis and multivariate regression for quantitative analysis. The multivariate regression model consists of building a relationship between a desired physical, chemical or biological attribute of an object and its spectra.