Multivariate Classification of Cuban Aged Rums

Abstract

Several instrumental analytical techniques that are employed in food analysis produce a large amountof data. The employment of the multivariate analysis allows extracting the most important informationfrom these data efficiently. Supervised classification is primarily used to build classification rules fora number of known subgroups. New samples are then assigned to the most likely subgroup based onthese rules. The present work had as objective the application of two techniques of supervisedclassification, modeling independent class analysis (SIMCA) and partial least square discriminantanalysis (PLS-DA) to data of gas chromatography and acidity for 52 samples of aged Cuban rum, toachieve the dark aged rum classification among several aged rum analyzed in the Reference center ofdrinks and alcohols (CERALBE) of the Instituto Cubano de Investigaciones de los Derivados de laCaña de Azúcar (ICIDCA). The software UNSCRAMBLER v. 8.0 (CAMO ÅS, N-7401) was used. Themore important result obtained is that 100% of the samples of aged dark rum was very well classifiedby the SIMCA and by PLS-DA. The advantages and disadvantages of both methods are analyzed aswell as the future use of these models in the industry.