“…This dimensionality reduction technique takes an m-dimensional cloud of points, where m is the number of wavelengths in the spectral data, and transforms it so that in the new orthogonal coordinate space the first coordinate (first principal component, PC1) corresponds to the direction in which the original cloud showed the greatest variance, the second coordinate (PC2) represents the direction with the second greatest variance, and so on. The new coordinate system usually does not have an exact physical meaning, but its usefulness has been proven in previous research: for example, for asteroid taxonomy based on PCA, see DeMeo et al (2009); for analysis of meteorite spectra, see Penttilä et al (2018); for research on the lunar magnetic anomalies, see Chrbolková et al (2019) and Kramer et al (2011); and for PCA applied to the spectra of galaxies, see Connolly & VanderPlas (2014).…”