Multi-parameter prediction of oil palm fruit quality through near infrared spectroscopy combined with chemometric...

Early information concerning water, oil, and free fatty acid (FFA) content in palm fruit on-site is crucial in determining the commercial value of fresh fruit bunches (FFB) and maintaining oil palm quality. Conventional methods are destructive, labor-intensive, single-target, costly, and time-consuming. Therefore, this study aimed to develop a method for the multi-parameter prediction of oil palm fruit quality using near-infrared (NIR) spectroscopy. Empirical Wavelet Transform (EWT) and Gaussian Process Regression (GPR) were implemented as a novel chemometric method. A total of 750 fruit samples of the Tenera variety (Elaeis guineensis Jacq. var. tenera) with various maturity levels were collected from the Cikabayan Oil Palm Plantation. Each sample was scanned using an NIR instrument at 1000-1500 nm wavelengths to obtain absorbance data. The EWT was applied to decompose the NIR spectra into empirical modes, and GPR was used to build a regression model for fitting. Based on numerical analysis, the combination of EWT and GPR produced the root mean square error (RMSE) values of 2.877 ± 0.900 % (R² = 0.955 ± 0.018), 1.256 ± 0.543 % (R² = 0.942 ± 0.030), and 0.065 ± 0.04 % (R² = 0.964 ± 0.044) for water, oil, and FFA content, respectively. The results showed that the model could accurately predict the internal quality of oil palm fruit without the need for solvents or reagents, supporting environmental sustainability. These were expected to enhance oil palm production management by improving quality control, optimizing harvest timing, and promoting sustainability across the value chain.