A Vis-NIRs Calibration Model for the Prediction of Myristicin and Alpha-Pinene on Nutmeg: A Comparison Study of PLSR Algorithm and Machine Learning Algorithm

Determination of the myristicin and alpha-pinene content nutmeg is still constrained by extended testing time in laboratory, which expensive carried out destructively. In addition, non-destructive using spectroscopy often faces problems building models that only rely on algorithms perform linearly, such as PCR PLSR. Therefore, present study studied Vis-NIR (381-1065 nm) a fast, inexpensive, mechanism to determine fruits from Aceh, Indonesia. Two commonly used spectral data processing, partial least squares regression (PLSR) machine learning represented support vector (SVM), were employed compared predict fruits. The chemical reference parameters (myristicin alpha-pinene) measured gas chromatography mass spectrometry (GC-MS). Standard normal variate (SNV) multiplicative scatter correction (MSC) preprocessing involved spectra enhancement before prediction outcome. results show kernel radial basis function (RBF) n-SVM algorithm better than PLSR for with gamma (g), c, nu (n) 0.1, 1.0, 0.99, respectively. Also, e-SVM RBF epsilon (e) compositions 0.01, 10, coefficient correlation calibration (rc) determination (Rp2), root means square error (RMSEC) (RMSEP), ratio (RPD) 0.992, 0.986, 0.941%, 1.325% 8.348, mean deviation 0.976, 0.979, 0.305%, 0.317% 6.826, general, satisfactorily indicate Vis-NIRS, appropriate algorithm, has promising determining Indonesia, nondestructive measurement.