Improving the accuracy of near-infrared (NIR) spectroscopy method to predict the oil content of oil palm fresh fruits

Abstract

Abstract Oil content is an important factor to determine the price of the oil palm fruit. In this study, the NIR Spectroscopy method was conducted to determine the oil content of oil palm fresh fruits. Several studies that related to the NIR spectroscopy method for predicting the oil content of oil palm fruits showed that the accuracy was still optimal yet. This study is the initial stage to develop the optical portable instrument that can predict the oil content of oil palm fresh fruits. Five hundred samples which are divided into ten groups based on maturity levels were prepared for reflectance and oil content chemical measurement. The reflectance of the sample was measured by the FT-NIR spectrometer in the wavelength of 1000-1500 nm. The obtained spectrum of oil palm fresh fruits was transformed to absorbance (Log 1/R) and several spectra data processing was conducted to increase the accuracy of prediction. The calibration and validation of processed NIR spectra and the oil content were conducted using Partial Least Square (PLS) and MLR methods. Generally, PLS and MLR methods can be used to improve the accuracy of the NIR spectroscopy method. Then, the result showed that the MLR method is less accurate than the PLS method to predict the oil content of oil palm fresh fruits. In this study, the best result could be determined by the PLS model using 5 factors and spectra data processing of the first derivative of spectra absorbance (R=0.879; CV=19.8%; RPD= 2.46). A lower accuracy was obtained by MLR model (R=0.677; CV=28.33%; RPD =1.72).