An approach to the generation of synthetic Raman spectra using the principal component method is proposed. The Raman spectra obtained for adipose tissue before and after lipase exposure are used as initial data. The proposed data are characterized by the similarity of the spectra, since the studied materials have a similar chemical structure: adipose tissue consists of triglyceride molecules. After exposure to lipase, a mixture of di- and monoglycerides, as well as free fatty acids, is formed. The approach proposed in the work to generate spectral data is as follows: (1) applying the principal component analysis (PCA) to a set of real spectral data, (2) calculating the average value and standard deviation of the scores of a selected number of principal components (PC), (3) generating a new set of PC scores based on the average value and standard deviation, (4) generating spectra according to the PC loadings of the selected number of main components and the new set of PC scores. To confirm the similarity of the initial and synthetic data, a random forest classification model was trained on the initial data, followed by an accuracy assessment on the synthetic data. The proposed approach made it possible to perform augmentation process of Raman spectroscopy data with an average accuracy of 90.6± 0.5%. Keywords: Raman spectroscopy, data augmentation, principal component analysis, machine learning, random forest.