Coefficient Of Determination In Calibration Research Articles

Application of fecal near-infrared reflectance spectroscopy profiling for the prediction of diet nutritional characteristics and voluntary intake in beef cattle.

The objective of this study was to evaluate the use of fecal near-infrared reflectance spectroscopy (NIRS) profiling to predict diet nutritional characteristics and voluntary DMI in beef cattle. Fecal samples were collected for growing cattle across 11 experiments in which individual animal performance and DMI was measured. Dried and ground fecal composite samples collected from each animal were subjected to fecal NIRS analysis by a Foss NIRS 6500 scanning monochromator (Foss, Eden Prairie, MN) at the Grazingland Animal Nutrition Laboratory (Temple, TX). Fecal spectra were then used to develop equations to predict diet composition (trials 1 to 11; = 408), digestibility (trials 1 to 5; = 155), and DMI (trials 1 to 11; = 408). Coefficients of determination for calibration () and cross-validation () for prediction of diet nutritional characteristics were lower for NDF ( = 0.85; = 0.82) than for CP ( = 0.90; = 0.88). For the prediction of DMI, and ranged from 0.69 and 0.67 for the prediction of trial-average DMI to 0.76 and 0.73 for the prediction of fecal-collection-period DMI. While the and obtained for the prediction of DMI were lower than those obtained for the prediction of diet composition or digestibility, fecal NIRS prediction equations for DMI were successful in predicting the mean DMI of groups, as no differences were found for the prediction of fecal-collection-period DMI (Diff. = 1.10; = 0.72) or trial DMI (Diff. = -0.47; = 0.86).

Near-infrared reflectance spectroscopy for the prediction of chemical composition in walnut kernel

ABSTRACTIn the present work, 116 samples were collected and near-infrared reflectance spectroscopy prediction model for determination of moisture, protein, and fat contents of walnut meal were obtained and evaluated. All the samples were analyzed based on the chemical methods. Meanwhile, they were scanned to obtain their near-infrared reflectance spectrum in the wavelength range of 570–1840 nm. Several preprocess treatments including scattering pretreatments, mathematical pretreatments, and aggression methods were optimized to increase the accuracy of the calibration models according to the coefficient of determination for calibration (Rc2) and the cross-validation (one minus the variance ratio, 1-VR), and the standard error of calibration and cross-validation. The results showed modified partial least square loading was the better aggression method to predict the moisture, proteins, and fats in walnut kernel. The calibration equations obtained indicated that near-infrared reflectance spectroscopy had excellent predictive capacity for the three components with Rc2 = 0.965, standard error of calibration = 0.052 for moisture, and Rc2 = 0.967, standard error of calibration = 0.191 for proteins, and Rc2 = 0.979, standard error of calibration = 0.171 for fats, respectively. The external validation further confirmed the robustness and reliability of the near-infrared reflectance spectroscopy prediction models with the correlation coefficient of actual and predicted values (R2) = 0.952, ratio of performance deviation = 4.14, the standard error of prediction =0.058 for moisture, and R2 = 0.977, ratio of performance deviation = 5.55, standard error of prediction = 0.182 for proteins, and R2 = 0.990, ratio of performance deviation = 8.64, standard error of prediction = 0.191 for fats, respectively. Near-infrared reflectance spectroscopy is a reliable technology to predict these constituents in walnuts.

Determination by near infrared microscopy of the nitrogen and carbon content of tomato (Solanum lycopersicum L.) leaf powder.

Near infrared microscopy (NIRM) has been developed as a rapid technique to predict the chemical composition of foods, reduce analytical costs and time and ease sample preparation. In this study, NIRM has been evaluated as an alternative to classical chemical analysis to determine the nitrogen and carbon content of small samples of tomato (Solanum lycopersicum L.) leaf powder. Near infrared spectra were obtained by NIRM for independent leaf samples collected on 216 plants grown under six different levels of nitrogen. From these, 30 calibration and 30 validation samples covering the spectral range of the whole set were selected and their nitrogen and carbon contents were determined by a reference method. The calibration model obtained for nitrogen content proved to be excellent, with a coefficient of determination in calibration (R2c) higher than 0.9 and a ratio of performance to deviation (RPDc) higher than 3. Statistical indicators of prediction using the validation set were also very high (R2p values > 0.90). However, the calibration model obtained for carbon content was much less satisfactory (R2c < 0.50). NIRM appears as a promising and suitable tool for a rapid, non-destructive and reliable determination of nitrogen content of tiny samples of tomato leaf powder.

Use of near-infrared reflectance spectroscopy for the rapid determination of the digestible energy and metabolizable energy content of corn fed to growing pigs.

BackgroundThe ability of near-infrared reflectance spectroscopy (NIRS) to determine the digestible energy (DE) and metabolizable energy (ME) content of corn fed to growing pigs was tested. One hundred and seventeen corn samples, comprising different planting regions and varieties were collected from all over China in a three-year period. The samples were randomly split into a calibration set (n = 88) and a validation set (n = 29). The actual and calculated DE and ME content of the corn samples was determined by digestion-metabolism experiments and the prediction equations of Noblet and Perez (J Anim Sci. 71:3389–98,1993). The samples were then subjected to NIRS scanning and calibrations were performed by the modified partial least square (MPLS) regression method based on 77 different spectral pre-treatments. The NIRS equations based on the actually determined and calculated DE and ME were built separately and then validated using validation samples.ResultsThe NIRS equations obtained from actually determined DE, the coefficient of determination for calibration (RSQcal), cross-validation (R2CV), and validation (RSQv) were 0.89, 0.87 and 0.86, and these values for determined ME were 0.87, 0.86 and 0.86. For the NIRS equations built from calculated DE, the RSQcal, R2CV, and RSQv values were 0.88, 0.85 and 0.84, and these values for calculated ME were 0.86, 0.84 and 0.82. Except for the equation based on calculated ME (RPDv = 2.38, < 2.50), the other three equations built from actually determined energy and calculated DE produced good prediction performance (RPDv ranging from 2.53 to 2.69, > 2.50) when applied to validation samples.ConclusionThese results indicate that NIRS can be used as a quantitative method for the rapid determination of the available energy in corn fed to growing pigs, and the NIRS equations based on the actually determined energy produced better predictive performance than those built from calculated energy values.

Rapid determination of chemical composition and classification of bamboo fractions using visible-near infrared spectroscopy coupled with multivariate data analysis.

BackgroundDuring conversion of bamboo into biofuels and chemicals, it is necessary to efficiently predict the chemical composition and digestibility of biomass. However, traditional methods for determination of lignocellulosic biomass composition are expensive and time consuming. In this work, a novel and fast method for quantitative and qualitative analysis of chemical composition and enzymatic digestibilities of juvenile bamboo and mature bamboo fractions (bamboo green, bamboo timber, bamboo yellow, bamboo node, and bamboo branch) using visible–near infrared spectra was evaluated.ResultsThe developed partial least squares models yielded coefficients of determination in calibration of 0.88, 0.94, and 0.96, for cellulose, xylan, and lignin of bamboo fractions in raw spectra, respectively. After visible–near infrared spectra being pretreated, the corresponding coefficients of determination in calibration yielded by the developed partial least squares models are 0.994, 0.990, and 0.996, respectively. The score plots of principal component analysis of mature bamboo, juvenile bamboo, and different fractions of mature bamboo were obviously distinguished in raw spectra. Based on partial least squares discriminant analysis, the classification accuracies of mature bamboo, juvenile bamboo, and different fractions of bamboo (bamboo green, bamboo timber, bamboo yellow, and bamboo branch) all reached 100 %. In addition, high accuracies of evaluation of the enzymatic digestibilities of bamboo fractions after pretreatment with aqueous ammonia were also observed.ConclusionsThe results showed the potential of visible–near infrared spectroscopy in combination with multivariate analysis in efficiently analyzing the chemical composition and hydrolysabilities of lignocellulosic biomass, such as bamboo fractions.Electronic supplementary materialThe online version of this article (doi:10.1186/s13068-016-0443-z) contains supplementary material, which is available to authorized users.

Prediction of Beef Fat Content Simultaneously under Static and Motion Conditions Using near Infrared Spectroscopy

Fat content is one of the most important quality indicators for minced beef products. In this study, a multipoint near infrared (NIR) spectrophotometer system, based on a Fabry–Perot interferometer, combined with a four-point photodiode array detector and flexible collimator–probe arrangement, was used for real-time analysis of beef fat content. The system was employed to predict fat content of mixed minced beef samples concurrently under two different conditions: (a) static and slow motion and (b) static and fast motion. Additionally, a separate measurement was conducted to further test the independency of a collimator–probe arrangement by scanning two samples with different fat percentages concurrently under static and motion conditions. Partial least squares regression was employed, obtaining coefficients of determination in calibration ( R2c) of 0.95, confirming a good fit for the three models. The fat contents of samples in the independent set were predicted with reasonable accuracy: r2 in the range 0.82–0.92 and standard error of prediction in the range 3.05–3.98%. Moreover, the spectral features observed for the probe independency test clearly illustrated the flexibility and independency of the collimator–probe arrangement. This study showed that the multipoint NIR spectroscopy system can predict beef fat content concurrently under static and motion conditions and illustrates its potential use as an in-line monitoring tool at various junctions in a meat processing plant.

NIR assignment of isopsoralen by 2D-COS technology and model application in Yunkang Oral Liquid

Near infrared (NIR) assignment of Isopsoralen was performed using deuterated chloroform solvent and two-dimensional correlation spectroscopy (2D-COS) technology. Yunkang Oral Liquid was applied to study Isopsoralen, the characteristic bands by spectral assignment as well as the bands by interval partial least squares (iPLS) and synergy interval partial least squares (siPLS) were used to establish partial least squares (PLS) model. The coefficient of determination in calibration [Formula: see text] were 0.9987, 0.9970 and 0.9982. The coefficient of determination in cross validation [Formula: see text] were 0.9985, 0.9921 and 0.9982. The coefficient of determination in prediction [Formula: see text] were 0.9987, 0.9955 and 0.9988. The root mean square error of calibration (RMSEC) were 0.27, 0.40 and 0.31 ppm. The root mean square error of cross validation (RMSECV) were 0.30, 0.67 and 0.32 ppm. The root mean square error of prediction (RMSEP) were 0.23, 0.43 and 0.22 ppm. The residual predictive deviation (RPD) were 31.00, 16.58 and 32.41. It turned out that the characteristic bands by spectral assignment had the same results with the chemometrics methods in PLS model. It provided guidance for NIR spectral assignment of chemical compositions in Chinese Materia Medica (CMM).

Optimization of Parameter Selection for Partial Least Squares Model Development.

In multivariate calibration using a spectral dataset, it is difficult to optimize nonsystematic parameters in a quantitative model, i.e., spectral pretreatment, latent factors and variable selection. In this study, we describe a novel and systematic approach that uses a processing trajectory to select three parameters including different spectral pretreatments, variable importance in the projection (VIP) for variable selection and latent factors in the Partial Least-Square (PLS) model. The root mean square errors of calibration (RMSEC), the root mean square errors of prediction (RMSEP), the ratio of standard error of prediction to standard deviation (RPD), and the determination coefficient of calibration (Rcal2) and validation (Rpre2) were simultaneously assessed to optimize the best modeling path. We used three different near-infrared (NIR) datasets, which illustrated that there was more than one modeling path to ensure good modeling. The PLS model optimizes modeling parameters step-by-step, but the robust model described here demonstrates better efficiency than other published papers.

Rapid Determination of Lycium Barbarum Polysaccharide with Effective Wavelength Selection Using Near-Infrared Diffuse Reflectance Spectroscopy

A rapid method to evaluate the amount of Lycium barbarum polysaccharides (LBPs) based on near-infrared diffuse reflectance spectroscopy (NIDRS) is proposed. A total of 114 dried L. barbarum samples from different regions of China were used for modeling by partial least squares (PLS) regression with the effective wavelengths (EWs) selected by correlation coefficients and X-loading weights. The PLS model worked best with EWs of 4003–5087, 5568–7002, and 7463–12,000 cm−1. These EWs were chosen by correlation coefficients after spectral pretreatment. Use of these EWs resulted in determination coefficients of calibration and validation of 0.982 and 0.938, respectively. The proposed model successfully predicted external samples, which further confirmed prediction ability of the model. Overall results indicated that both correlation coefficients and X-loading weights for EWs selection can improve model performance and that NIDRS can rapidly determine LBPs with intact particle samples.

Prediction of Starch Content and Ethanol Yields of Sorghum Grain Using near Infrared Spectroscopy

A rapid quantification method was developed and validated for non-destructive measurement of starch content, theoretical ethanol yield and actual ethanol yield of 48 cultivars of sorghum grain using Fourier transform near infrared (FT-NIR) spectroscopy in diffuse reflectance mode. Multiplicative scatter correction, Savitzky–Golay derivative smoothing and mean centring were used for processing the spectra of ground sorghum grain. The processed spectra were correlated with starch content, theoretical ethanol yield and ethanol produced through simultaneous saccharification and fermentation using partial least-squares regression (PLSR). The spectral range and number of factors were optimised for the low number of factors, high coefficients of determination for calibration ( R2) and validation ( r2), low root mean square error of prediction ( RMSEP), high ratio of performance to deviation ( RPD) and high ratio of the standard error of prediction to the range ( RER). The best PLSR model for starch content utilised the 4000–6000 cm−1 wavebands and had the following values: R2 = r2 = 0.97, RMSEP = 5.5 g kg−1 grain, RPD = 5.9 and RER = 15. Likewise, the model for theoretical ethanol yield utilised the 4000–8000 cm−1 wavebands and had R2 and r2 values of &gt;0.90, RMSEP = 4.9 g kg−1 grain, RPD = 4.47 and RER = 12.8. It was more difficult to predict actual ethanol yield using FT-NIR spectroscopy given the small data set, and spectra were collected prior to the fermentation step. Resulting PLSR models had R2 and r2 values of &lt;0.60, RMSEP = 11.2–21.4 g kg−1, RPD &lt; 3 and RER &lt; 6. These results demonstrated that FT-NIR spectroscopy may be a practical method for rough screening of sorghum cultivars for desirable starch content and theoretical ethanol yield. The models may be improved by including more cultivars in the model and additional compositional information, such as tannin and free amino nitrogen contents, in the chemometric analysis and using FT-NIR scans of the fermentation products to predict actual ethanol yields.

Determination of the degree of degradation of frying rapeseed oil using fourier-transform infrared spectroscopy combined with partial least-squares regression.

This rapid method for determining the degree of degradation of frying rapeseed oils uses Fourier-transform infrared (FTIR) spectroscopy combined with partial least-squares (PLS) regression. One hundred and fifty-six frying oil samples that degraded to different degrees by frying potatoes were scanned by an FTIR spectrometer using attenuated total reflectance (ATR). PLS regression with full cross validation was used for the prediction of acid value (AV) and total polar compounds (TPC) based on raw, first, and second derivative FTIR spectra (4000–650 cm−1). The precise calibration model based on the second derivative FTIR spectra shows that the coefficients of determination for calibration (R 2) and standard errors of cross validation (SECV) were 0.99 and 0.16 mg KOH/g and 0.98 and 1.17% for AV and TPC, respectively. The accuracy of the calibration model, tested using the validation set, yielded standard errors of prediction (SEP) of 0.16 mg KOH/g and 1.10% for AV and TPC, respectively. Therefore, the degradation of frying oils can be accurately measured using FTIR spectroscopy combined with PLS regression.

Determination of the Carbonyl Values for Frying Rapeseed Oil Using Near-Infrared Spectroscopy

Near-infrared (NIR) spectroscopy was investigated as a method for analysis of the carbonyl value (CV) in frying oils. Two types of rapeseed oil were used to fry potatoes and 156 samples of frying oil were obtained. Transmission spectra of the samples were acquired directly using glass test tubes (13-mm diameter) as sample cells and an NIR spectrometer. Calibration models with very high accuracy based on the second derivative spectra in wavelength range of 1100–2200 nm were developed for predicting the CV using partial least-squares regression with full cross-validation. The coefficients of determination for calibration and standard error of cross-validation were 0.98 and 1.79 μmol g−1, respectively. The accuracy of the NIR calibration models was tested using the validation set, yielding the value for the root mean square of the prediction as 1.74 μmol g−1. The results demonstrate that the CV for frying oils can be accurately determined using NIR spectroscopy.

Development, validation and influence factor analysis of a near-infrared method for the molecular weight determination of xanthan gum

A practical molecular weight determination model of xanthan gum (XG), based on near-infrared (NIR) spectroscopy, was built in this study. Two sample measurement modules, integrating sphere module and fiber-optic probe module, were compared, and the best partial least square (PLS) regression model was based on fiber-optic probe module. The values of coefficient of determination in calibration (R2c), coefficient of determination in prediction (R2p), residual predictive deviation (RPD) and root mean square error of prediction (RMSEP) were 0.967, 0.975, 6.028 and 0.250×106Da, respectively. The molecular weight range, linearity, accuracy and precision of the established method were also validated. Furthermore, influence factors on this method were discussed in order to establish an appropriate measurement protocol. Results showed that the proposed NIR method may be suitable for practical applications in manufacturing plants and probably be accepted as a good alternative approach for fast determination of molecular weight of XG in production process.

Development of near infrared reflectance spectroscopy to predict chemical composition with a wide range of variability in beef

A total of 182 beef samples were minced and divided into calibration set (n=140) and independent validation set (n=42). Calibration models of NIRS (1000–1800nm) were built using partial least squares regression (PLSR) on the calibration set of samples. Both the coefficient of determination in calibration (R2C) and the coefficient of determination in prediction (R2P) were over 0.98 for all chemical compositions. The ratio performance deviation (RPD) was 17.37, 5.12 and 10.43 for fat, protein and moisture, respectively. The results of the present study indicate the outstanding ability of NIRS to predict chemical composition in beef.

근적외 분광분석법을 이용한 국내 유통 식품 함유 탄수화물, 단백질 및 지방의 정량 분석

Foods contain various nutrients such as carbohydrates, protein, oil, vitamins, and minerals. Among them, carbohydrates, protein, and oil are the main constituents of foods. Usually, these constituents are analyzed by the Kjeldahl and Soxhlet method and so on. However, these analytical methods are complex, costly, and time-consuming. Thus, this study aimed to rapidly and effectively analyze carbohydrate, protein, and oil contents with near-infrared reflectance spectroscopy (NIRS). A total of 517 food samples were measured within the wavelength range of 400 to 2,500 nm. Exactly 412 food calibration samples and 162 validation samples were used for NIRS equation development and validation, respectively. In the NIRS equation of carbohydrates, the most accurate equation was obtained under 1, 4, 5, 1 (1st derivative, 4 nm gap, 5 points smoothing, and 1 point second smoothing) math treatment conditions using the weighted MSC (multiplicative scatter correction) scatter correction method with MPLS (modified partial least square) regression. In the case of protein and oil, the best equation were obtained under 2, 5, 5, 3 and 1, 1, 1, 1 conditions, respectively, using standard MSC and standard normal variate only scatter correction methods with MPLS regression. Calibrations of these NIRS equations showed a very high coefficient of determination in calibration (R²: carbohydrates, 0.971; protein, 0.974; oil, 0.937) and low standard error of calibration (carbohydrates, 4.066; protein, 1.080; oil, 1.890). Optimal equation conditions were applied to a validation set of 162 samples. Validation results of these NIRS equations showed a very high coefficient of determination in prediction (r²: carbohydrates, 0.987; protein, 0.970; oil, 0.947) and low standard error of prediction (carbohydrates, 2.515; protein, 1.144; oil, 1.370). Therefore, these NIRS equations can be applicable for determination of carbohydrates, proteins, and oil contents in various foods.

The application of near infrared spectroscopy to study lipid characteristics and deterioration of frozen lean fish muscles

Near infrared spectroscopy (NIR) was applied to estimate lipid composition and degradation of two lean fish species, saithe (Pollachius virens) and hoki (Macruronus novaezelandiae). Calibration models were developed, using partial least squares (PLS) regression, for total lipid content and composition, free fatty acids (FFA), thiobarbituric acid reactive substances (TBARS) and fluorescent interaction compounds (OFR). Coefficients of determination for calibration (R2cv) and root-mean-square error of cross validation (RMSECV) ranged from 0.82 to 0.99 and 0.66 to 3.69 for hoki and from 0.64 to 0.99 and 0.06 to 2.65 for saithe, respectively. The validations of the calibrations indicated that lipid composition and FFA of hoki and saithe can be estimated by NIR with good accuracy. Furthermore, NIR differentiate fish muscles with low, medium and high concentration of OFR and TBARS. Overall, the results demonstrate the potential for use of NIR spectroscopy as an objective and non-destructive method to inspect the lipid characteristics and quality of frozen lean fish.

In Situ Measurement of Soil Chemical Composition by Near-Infrared Spectroscopy: A Tool Toward Sustainable Vineyard Management

This study evaluated an in-field near-infrared (NIR) instrument to predict the contents of total nitrogen (TN), organic carbon (OC), potassium (K), sulfur (S), phosphorus (P), pH, and electric conductivity (EC) in soil vineyard samples (n = 70) sourced from three wine regions of Australia. Samples were analyzed using a portable NIR spectrophotometer (ASD FieldSpec III, 350–1800 nm). Partial least squares (PLS) regressions yield a coefficient of determination in calibration (R2) and a standard error in cross validation (SECV) of 0.74 (0.03) for TN, 0.92 (2.19) for S, 0.81 (0.42) for OC, 0.70 (109.2) for K, 0.84 (0.03) for EC, 0.83 (0.44) for pH, and 0.69 (24.6) for P, respectively. This study showed that it is possible to measure soil chemical properties in the vineyard, and the main advantages of this approach will be the speed, low cost, and ability to better manage and monitor soil fertility.

Assessing the Value of a Portable Near Infrared Spectroscopy Sensor for Predicting Pork Meat Quality Traits of “Asturcelta Autochthonous Swine Breed”

Sixty-one intact meat samples from Asturcelta autochthonous swine breed were scanned in the slaughterhouse in reflectance mode. A handheld microelectromechanical system digital transform (Phazir1624, Polychromix Inc.), with a window sampling area of 0.8 × 1 cm and wavelengths ranging from 1,600 to 2,400 nm, was used. With the spectra database recorded were developed different chemometrical models assaying first and second derivatives as math treatment and standard normal variate (SNV) and multiplicative scatter correction for minimizing scattering effect. The greatest predictive capacity was achieved after applying SNV and first derivative for moisture, intramuscular fat (IMF) content, and pH parameters and second derivative for CIE L*, a*, b* colorimetric values, and the Warner–Bratzler force (instrumental texture). The coefficients of determination for calibration ranged from 0.63 to 0.89. The ratio between the standard error of the laboratory and the standard error of calibration ranged from 0.8 to 2.5 for all parameters (1.7 on average) with the exception of b and pH with ratios of 3.5 and 4.1, respectively. The statistical values obtained for the models developed to estimate IMF, CIE L*, a*, b*, moisture, and pH, displayed acceptable predictive capacity. For instrumental texture, the model could be able to discriminate among tender, medium, and hard meat in carcasses for characterization slaughter purposes.

Evaluation of potential nirs to predict pastures nutritive value

Abstract This paper describes the capability of near infra-reflectance (NIRS) to predict the nutritional quality of pastures from southern Chile (39°-40°S). A Fourier transformed near-infrared (FT-NIR) method for rapid determination of dry matter (DM), crude protein (CP), in vitro digestibility (IVD) and metabolizable energy (ME) was used. Calibration models were developed between chemical and NIRS spectral data using partial least squares (PLS) regression and external validation. The coefficients of determination in calibration (R 2 c ) were high varying between 0.89-0.99 and the root mean square errors of calibration (RMSEC) were low, ranging between 0.46-2.55 for the parameters analysed. The Residual Prediction Deviation (RPD) was higher than 2.5. Our results confirmed the convenience of using a wide range of samples applicability in the calibration set. Data also showed that the use of an independent set of samples for external validation increases the robustness of the models to predict unknown samples. Our results indicated RPD values higher than 2.5 which is the minimum recommended for this type of prediction. Thus, the result showed that NIRS was useful to estimate the nutritional quality of permanent pastures, and has a great potential to be used as a rapid decision tool for the studied analysis.

Oil accumulation in intact olive fruits measured by near infrared spectroscopy–acousto‐optically tunable filter

A field experiment was conducted to test the reliability of the near infrared spectroscopy (NIR)-acousto-optically tunable filter (AOTF) method to measure mesocarp oil content in vivo against nuclear magnetic resonance (NMR) determinations using three different olive cultivars at different stages of ripening. In the partial least squares model carried out for the cultivar 'Arbequina', the coefficient of determination in calibration (R(2)c) was 0.991, while the coefficient of determination in cross-validation (R(2)cv) was 0.979. For the cultivar 'Frantoio' the indexes were 0.982 and 0.971, respectively; while for the cultivar 'Leccino' R(2)c was 0.977 and R(2)cv was 0.965. Finally, for the combined model (sum of the three varieties) these indexes were 0.921 and 0.903, respectively. The residual predictive deviation (RPD) ratio was insufficient for the predictive model of cultivar 'Leccino' only (1.98), whereas in the other cases the RPD ratios were completely sufficient, within the estimation range over 2.5-3 (2.61 in the global model, and 4.23 in the cultivar 'Frantoio'), or in describing a large capacity with values greater than 5, as in the cultivar 'Arbequina' (9.58). NIR-AOTF spectroscopy proved to be a novel, rapid and reliable method to monitor the oil accumulation process in intact olive fruits in the field. The innovative approach of coupling NIR and NMR technologies opens up new scenarios for determining the optimal time for harvesting olive trees to obtain maximum oil production.