Prediction of Fermentation Index and pH of Cocoa (Theobroma cacao L.) Beans Based on Color Features (Cut Test) and Partial Least Square Regression Model

Abstract

Post-harvest processes such as the fermentation stage are important parameters to determine the quality and price of cocoa beans. The color change of cocoa beans from dark purple to brown is an indicator of success in the cocoa fermentation process. So far, farmers use an estimation system and use laboratory equipment which has many shortcomings and is difficult to apply to farmer groups. The application of technology to measure the rate of fermentation quickly and accurately encourages several scientists to build a predictive model that is used as the initial stage of making a technology. The purpose of this study was to determine the prediction results of the Fermentation Index and pH of cocoa beans using a partial least square regression model and to determine the relationship between the image of cocoa beans (cutting test) with the pH and fermentation index of cocoa beans. The method uses primary data from research in the laboratory for pH and cocoa fermentation index measurements and cocoa bean image processing to develop a predictive model. The result of this study is a partial least square regression model for predicting the cocoa bean fermentation index with cross validation 5 times shown that, partial least square regression model for pH prediction shows an accuracy of 99% and the error value is only 0.007.