Progrès dans la science et la technologie des cultures

Abstrait

Prediction of Fat Content in Intact Cocoa Beans Using Near Infrared Reflectance Spectroscopy

Zulfahrizal, Sutrisno, Seminar KB, Munawar AA and Budiastra IW

The majority of Indonesia’s cocoa export is raw beans which accounted by 82% of the total export. Indonesian cocoa beans are only used as additional material by cocoa industrialized countries due to low quality. The objective of this research was to study and analyze the application of near infrared reflectance spectroscopy (NIRS) method coupled with partial least squares (PLS) to determine the quality of cocoa particularly to predict the fat content in intact cocoa beans which has never been conducted before. Besides, this research was also to study the application of six spectra pre-processing methods i.e. mean centering (MC), multiplicative scatter correction (MSC), standard normal variate (SNV), mean normalization (MN), orthogonal signal correlation (OSC) and de-trending (DT) in increasing the performance of PLS. It is found in this study that PLS combined with MSC and SNV provide prediction value with root mean square error calibration (RMSEC) were 0.93% and 0.91% respectively, whilst root mean square error prediction (RMSEP) and ratio of performance to deviation (RPD), generated from both spectra pre-processing were similar i.e. 1.11% and 1.95 respectively.