Understanding the structural diversity of nucleic acids, particularly RNA, is essential for elucidating their biological functions. However, classifying nucleic acid conformations using a single spectroscopic technique remains challenging due to their inherent structural variability and the absence of standardized spectral references for consistent comparison. This chapter presents a robust, data-driven workflow for defining and validating structural classes of nucleic acids based on spectral data. The method combines manual class definition, spectral normalization, dimensionality reduction via Singular Value Decomposition (SVD), and validation using both linear and nonlinear similarity metrics. An auto-iterative process ensures convergence and consistency in class assignment. This approach enables the creation of reliable reference spectra and supports the classification of unknown samples, offering a valuable tool for structural studies of nucleic acids across various spectroscopic techniques.