Prediction of chemical compositions of crushed maize used in meat animal ration using near infrared spectroscopy and multivariate analysis

Abstract

The aim of this study was to test the ability of near infrared (NIR) reflectance spectroscopy to predict the proximate components in crushed maize which is used in meat animal ration. In total 100 samples were prepared among these 75 samples were in calibration set and 25 samples were in validation set. For detection of chemical composition of crushed maize, spectra were collected using DLPNIR scan Nano Software. Partial least square regression (PLSR) model for calibration and validation were developed through The Unscrambler X software. Accuracies of the calibration models were evaluated using the root mean square error of calibration (RMSEc), root mean square error of validation (RMSEv), coefficient of calibration (R2 c) and coefficient of validation (R2 v). Generally, the accuracy (i.e. the closeness between actual and the predicted values) of regression model is considered as excellent when the R2 ≥ 0.90. Predictive ability of the PLSR model is assessed by co-efficient of determination of validation (R2V) and root mean square error of cross-validation (RMSEV). The best model for each trait is selected on the basis of the highest co-efficient of determination of validation (R2V) and the lowest root mean square error of validation (RMSEV). R2 v 0.972, 0.971, 0.971, 0.957, 0.960, 0.968 for DM, moisture, CP, EE, CF and ash respectively; and RMSEV values are 0.26, 0.26, 0.24, 0.15, 0.12, 0.19, for DM, moisture, CP, EE, CF and ash, respectively. From the findings it can be concluded that crushed maize proximate components can be predicted through PLS model from reference values using The Unscrambler X software.