The aim of this study was to compare the quality of a mixture of cocoa harvested and fermented in three subregions of Antioquia (Colombia), from the chemometric profile based on multivariate statistical analysis. A mixture of clones CCN-52, ICS-1, FLE-2, and FEC-2 harvested in Bajo Cauca, Uraba and Magdalena Medio were subjected to a spontaneous fermentation. The characterization of raw and well-fermented cocoa was performed through 38 parameters, and results were compared by a Principal Component Analysis (PCA) and a Cluster Analysis (CA), followed by a Principal Factors Analysis (PFA- CA). The CA showed that there are differences among subregions only in raw cocoa from Bajo Cauca. PCA allowed identifying the variability between raw and fermented cocoa in a representative way and these results were consistent with the chemical profile. Besides, the number of parameters to differentiate raw cocoa from different subregions was reduced (11-13 parameters) and it was possible to characterize well fermented cocoa with only 10 parameters of 38. PFA-CA consolidated in three factors a grouping to identify the cocoa quality according to the process or interest of the sensory or functional properties. Factor 1 (cocoa quality indicators with functional properties), Factor 2 (indicators of quality of the beginning of fermentation) and Factor 3 (indicators of quality of well-fermented cocoa) each one with a weight of 39, 35 and 26 respectively.
