Predictive modeling of color and sucrose in the cane sugar industry using key explanatory variables

The most influential variables affecting two key sugar quality characteristics sucrose content [percentage of sucrose] and ICUMSA color were identified, along with predictive models for each characteristic. The objective of this study was to determine the sugar production variables that have the greatest impact on sucrose content [% pol] and sugar color [°ICUMSA], and subsequently to develop predictive models capable of simulating different scenarios to support effective decision-making. Principal Component Analysis [PCA] was employed to extract the variables with the greatest influence on each quality characteristic. Subsequently, multiple linear regression was applied using the variables selected through PCA. As a result, two predictive models were obtained: one explaining 100% of the variability in sucrose content and another explaining 98% of the variability in ICUMSA color. Both models will be used to induce process changes that ensure compliance with the product quality specifications required by national and international customers.