Comparación de técnicas de procesamiento de datos obtenidos a través de espectros en el FTIR para la cuantificación simultánea de metanol y etanol presentes en bebidas alcohólicas artesanales producidas en el valle de Yunguilla

Abstract

This thesis addresses in detail the quantification of ethanol and methanol present in artisanal beverages produced in the Yunguilla region, using Fourier transform infrared spectroscopy (FTIR) as the main tool. The work focuses on applying a rigorous analytical approach using statistical and data processing programs, specifically through R Studio, which allowed for efficient management of experimental information. In the research, different predictive models were constructed using the sPLS multivariate regression methodology, a technique that is well suited to problems with high dimensionality of spectral variables. The study compared the use of the full spectrum with analysis focused solely on peaks of interest, allowing the advantages and limitations of each approach to be observed. Prior to modeling, the samples were preprocessed using the ALS and Savitzky-Golay methods. Finally, evaluation using the mean square error (MSE) criterion led to the conclusion that FTIR spectroscopy is a fast, practical, and technically viable alternative for the simultaneous quantification of ethanol and methanol.