The optical properties of minerals are determined by their chemical composition and structure as well as by diverse impurities and inclusions. Hence, spectroscopy is a powerful tool for the identification and discrimination of minerals. However, with these methods large sets of data are created in a short amount of time. As part of this work color coordinates of various gem stones were calculated using their Vis spectra and the CIE XYZ-System. This calculation reduces the data to only two values. However, a discrimination is only possible for strongly colored minerals. By means of multivariate methods information about subtle changes in the chemical composition and structure can be obtained even from apparently very similar spectra. The exploratory data analysis called principle component analysis (PCA) is used to achieve data reduction and make a clear presentation of the data possible. In this work calcite, magnesite, talc and quartz samples from different deposits were studied. Reflexion spectra in the Vis and NIR range were measured and a PCA model was calculated after appropriate pre-treatment of the data. In this model a formation of groups dependent on deposit location was observed for the magnesite, calcite and talc samples, but not for the quartz samples. In the NIR range 4 of 5 sample sets could be discriminated for calcite and magnesite. Additionally, 4 of 4 talc sample sets could be differentiated qualitatively. This discrimination was only partly possible in the Vis range.