Cross-validation Coefficient Of Determination Research Articles

Prediction of Density in Standing Trees of Various Wood Species in Natural Forests Using Near-infrared Spectroscopy

Density plays an important role as basic information for applying wood as construction materials. Recent years, the application of near-infrared spectroscopy as non-destructive testing (NDT) has been promising. Density prediction for standing trees in huge variation trees and species of natural forest needs to be investigated using NDT as of eco-green harvesting. The combination of density information and near-infrared spectroscopy is enabled to build a prediction model. This research applied increment cores sampling for density prediction analysis using near-infrared spectroscopy method. The research combined increment cores samples from multiple wood species to be analyzed in one chemometrics analysis of cross-validation partial least squares regression (CV-PLSR) to build a prediction model of density. The research resulted coefficient of determination for cross-validation (R2CV) of 0.76 with number of latent variable (LV) 10 from the 1st derivative with 13 smoothing-point spectra and wavelength of 1200 – 1800 nm as the best prediction model. The result seemed sufficient enough with those number of LV for this small tube wood sampling of increment cores from multiple wood species. This research proved that building a prediction model for multiple wood species is possible to be done.

Non-destructive prediction of ready-to-eat kiwifruit firmness based on Fourier transform near-infrared spectroscopy

There is a growing demand for ready-to-eat kiwifruit in the world. However, ready-to-eat kiwifruit has a rather narrow range of firmness (e.g. 10–30 N), and it remains challenging to predict this firmness in a non-destructive manner. Here, we report a strategy for non-destructive prediction of kiwifruit firmness based on Fourier transform near-infrared (FT-NIR) spectroscopy. The radial basis function (RBF) model displayed superior performance, with a coefficient of determination (Rc2) of 0.83, a cross-validation coefficient of determination (Rp2) of 0.73, a root mean square error of calibration (RMSEC) of 0.58, a root mean square error of prediction (RMSEP) of 0.72, and a ratio of performance to deviation (RPD) of 1.92. To enhance the accuracy of kiwifruit firmness prediction, we optimized the FT-NIR algorithm through data preprocessing, feature selection, and dimensionality reduction. The results showed that the FD-CARS-SVR (RBF) algorithm exhibited the best performance in predicting kiwifruit firmness during the shelf life with impressive values of Rc2 (0.99), Rp2 (0.92), RMSEC (0.15), RMSEP (0.40), and RPD (3.48). To further evaluate the applicability of the FT-NIR model, we compared the data predicted by the model and acquired from the Kiwifirm™ and penetrometer GY-4. The results revealed pronounced superiority of the FT-NIR model for the firmness ranging from 10 to 40 N to replace Kiwifirm™, providing a new non-destructive model for the prediction of the firmness of ready-to-eat kiwifruit.

Estimating leaf photosynthetic capacity using hyperspectral reflectance: Model variability and transferability

Leaf photosynthetic capacity is a crucial parameter for characterizing plant growth status and global nitrogen-carbon cycles. While leaf trait and reflectance models have been widely applied to assess leaf photosynthetic capacity across various plant species, the variability and transferability of these estimation models remain unclear. Thus, this study investigated the variability in estimating leaf maximum carboxylation rate of Rubisco (Vcmax, μmol m−2 s−1) using hyperspectral reflectance across seven plant species datasets. To improve model transferability, we proposed model updating by adding new samples and developed a stacking model to integrate multiple regression model results to reduce variability in the predictions. The PROSPECT model, coupled with spectral derivatives and similarity metrics, was used to retrieve leaf structural and biochemical traits. Our results showed that Vcmax was significantly correlated with the contents of leaf chlorophyll (Cab) and protein (Prot), and other traits such as leaf structure, carotenoid content, and water content also influenced Vcmax. However, the strength of these correlations varied among different datasets due to differences in vegetation types, growth periods, and the number of species. Leaf trait relationships also varied with datasets, with Cab proving to be a good proxy of photosynthesis across all datasets. Leaf traits were superior to leaf reflectance in characterizing the differences between datasets. While leaf reflectance performed well in estimating Vcmax for most datasets, leaf traits were more suitable for constructing transferable estimation models of Vcmax between different datasets. Model transferability was affected by differences in datasets, such as data range and plant species. Model updating by adding 10% new samples significantly improved the assessment of Vcmax, with leaf reflectance yielding better estimation results. Our data also revealed that different models produced inconsistent results, and a stacking model combining multiple models optimized the estimation of Vcmax using leaf traits and reflectance, with the cross-validation coefficient of determination and relative root mean square error of 0.88 and 19.92%, respectively. These findings offer new methods and ideas for assessing leaf photosynthetic capacity across different agro-ecosystems. Further recalibration of the proposed model with canopy radiative transfer models and datasets would enable monitoring of plant photosynthesis at large scales.

Application of near-infrared spectroscopy for predicting the wood density of teakwood stands by increment cores and thin wood surface sampling

Density, as the most critical information for wood materials could be predicted by near-infrared (NIR) spectroscopy. Increment cores and thin wood samples of teakwood ( Tectona grandis) served as research materials for this new approach to sampling for NIR spectroscopy, explained in detail. Density data were combined with their NIR spectra data for analysis using cross-validation partial least squares regression as chemometrics to produce a density prediction model. A new approach of increment cores from teakwood stands for NIR spectroscopy measurements for density predictions resulting in a 0.34 coefficient of determination for cross-validation (R2CV) and 0.40 of R2CV for thin wood samples as the best result from the first derivative with 25 smoothing points of NIR spectra. The improvement from the previous research for density prediction accuracy resulted in an R2CV value of 0.55. The smaller diameter of the increment core sample than the NIR light beam's diameter and illuminated mixed surface (transversal and radial) are supposed to give weaker prediction accuracy by this new approach.

Rapid assessment of residual solvent content in the TEGDN dual-base propellants by near-infrared reflectance spectroscopy

Excessive residual solvent (RS) levels in triethyleneglycol dinitrate (TEGDN) dual-base propellants can significantly impair combustion performance. This work aimed to develop a rapid and accurate model for detecting the RS content in the TEGDN dual-base propellants using near-infrared (NIR) spectroscopy in the reflectance mode. The optimal wavelength range for modelling, spanning from 1124.9−1230.2 nm and 1335.5–1527.5 nm, was identified based on absorption peaks characteristic of TEGDN dual-base propellant samples and RS. To enhance the quality of the data, we determined optimal window sizes for pre-processing methods: derivative pre-processing and Savitzky-Golay (S-G) smoothing pre-processing. After evaluating the performance of different pre-treatment methods, we found that the model employing multiple scattering corrections (MSC) in conjunction with first-order derivative (FD) pre-processing demonstrated superior results. The partial least squares (PLS) method was used to build the RS model with an optimal number of factors of 6. For the developed RS model, the root mean square error of calibration (RMSEC) and the root mean square error of cross-validation (RMSECV) were 0.019 and 0.024, respectively. The determination coefficient of calibration (Rc2) and the determination coefficient of cross-validation (Rcv2) were 0.968 and 0.952, respectively. In assessing the validation set using the developed model, we observed a root mean square error of prediction (RMSEP) of 0.025 and a determination coefficient of prediction (Rp2) of 0.958. Importantly, the relative error between the predicted values obtained through the NIR method and the measured values from the reference method consistently remained below 2 % under all circumstances. Consequently, the NIR-based RS model developed in this study offers a rapid and efficient means of detecting RS content in TEGDN dual-base propellants, facilitating judgment regarding the qualification of RS content.

The Use of a Droplet Collar Accessory Attached to a Portable near Infrared Instrument to Identify Methanol Contamination in Whisky.

The aim of this study was to evaluate the ability of a droplet collar accessory attached to a portable near-infrared (NIR) instrument to characterize the artificial contamination of methanol in commercial whisky samples. Unadulterated samples (n = 12) were purchased from local bottle shops where adulterated samples were created by adding methanol (99% pure methanol) at six levels (0.5%, 1%, 2%, 3%, 4% and 5% v/v) to the commercial whisky samples (controls). Samples were analyzed using a drop collar accessory attached to a MicroNIR Onsite instrument (900-1650 nm). Partial least squares (PLS) cross-validation statistics obtained for the prediction of all levels of methanol (from 0 to 5%) addition were considered adequate when the whole adulteration range was used, coefficient of determination in cross-validation (R2cv: 0.95) and standard error in cross of validation (SECV: 0.35% v/v). The cross-validation statistics were R2cv: 0.97, SECV: 0.28% v/v after the 0.5% and 1% v/v methanol addition was removed. These results showed the ability of using a new sample presentation attachment to a portable NIR instrument to analyze the adulteration of whisky with methanol. However, the low levels of methanol adulteration (0.5 and 1%) were not well predicted using the NIR method evaluated.

Development of NIRS calibration curves for sugars in baked sweetpotato.

Variability in sugar content between raw and cooked sweetpotato storage roots impact nutritional and dietary importance with implications for consumer preference. High-throughput phenotyping is required to breed varieties that satisfy consumer preferences. Near-infrared reflectance spectroscopy (NIRS) calibration curves were developed for analysing sugars in baked storage roots using 147 genotypes from a population segregating for sugar content and other traits. The NIRS prediction curves had high coefficients of determination in calibration (R2 c) of 0.96 (glucose), 0.93 (fructose), 0.96 (sucrose), and 0.96 (maltose). The corresponding coefficients of determination for cross-validation (R2 cv) were 0.92 (glucose), 0.89 (fructose), 0.96 (sucrose) and 0.93 (maltose) and were similar to the R2 c for all sugars measured. The ratios of the standard deviation of the reference set to the standard error of cross-validation were greater than three for all sugars. These results confirm the applicability of the NIRS curves in efficiently determining sugar content in baked sweetpotato storage roots. External validation was performed on an additional 70 genotypes. Coefficients of determination (r2) were 0.88 (glucose), 0.88 (fructose), 0.86 (sucrose) and 0.49 (maltose). The results were comparable to those found for the calibration and cross-validation in fructose, glucose, and sucrose, but were moderate for maltose due to the low variability of maltose content in the population. NIRS can be used for screening sugar content in baked sweetpotato storage roots in breeding programs and can be used to assist with the development of improved sweetpotato varieties that better meet consumer preferences. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Determination of the glass transition temperature of an epoxy prepreg by Near Infrared Spectroscopy

An aged epoxy prepreg's glass transition temperature (Tg) is determined using Near Infrared Spectroscopy (NIRS). A miniaturized Near Infrared (NIR) spectrometer and a full-scaled FTNIR spectrometer are used for Tg determination and different measuring setups are tested for the miniaturized spectrometer. Partial least square (PLS) regression is used to develop calibration models by utilizing different spectral pretreatments. It is shown that with the miniaturized spectrometer and a suitable measurement setup, PLS model qualities similar to those of the FTNIR spectrometer can be achieved in cross-validation (CV), with a determination coefficient of CV (R2CV) of about 0.97 and a root mean square error of cross-validation (RMSECV) of approx. 0.7 °C in the measuring range between 25 °C and 42 °C. Cost-effective, miniaturized NIR spectrometers are thus a suitable technology for the quality assurance of prepregs.

Application of near-infrared spectroscopy for the estimation of volatile compounds in Tempranillo Blanco grape berries during ripening.

The knowledge of volatile compounds concentration in grape berries is very valuable information for the winemaker, since these compounds are strongly involved in the final wine quality, and in consumer acceptance. In addition, it would allow to set the harvest date according to aromatic maturity, to classify grape berries according to their quality and to make wines with different characteristics, among other implications. However, so far, there are no tools that allow the volatile composition to be measured directly on intact berries, either in the vineyard or in the winery. In this work, the use of near-infrared (NIR) spectroscopy to estimate the aromatic composition and total soluble solids (TSS) of Tempranillo Blanco grape berries during ripening was evaluated. For this purpose, the spectra in the NIR range (1100-2100 nm) of 240 intact berry samples were acquired in the laboratory. From these same samples, the concentration of volatile compounds was analyzed by thin film-solid-phase microextraction-gas chromatography-mass spectrometry (TF-SPME-GC-MS), and the TSS were quantified by refractometry. These two methods were used as reference methods for model building. Calibration, cross-validation and prediction models were built from spectral data using partial least squares (PLS). Determination coefficients of cross-validation (R2 CV ) above 0.5 were obtained for all volatile compounds, their families, and TSS. These findings support that NIR spectroscopy can be successfully use to estimate the aromatic composition as well as the TSS of intact Tempranillo Blanco berries in a non-destructive, fast, and contactless form, allowing simultaneous determination of technological and aromatic maturities. © 2023 The Authors. Journal of The Science of Food and Agriculture published by John Wiley & Sons Ltd on behalf of Society of Chemical Industry.

Use of near-infrared reflectance spectroscopy on feces to estimate digestibility and dry matter intake of dietary nutritional characteristics under grazing conditions in Colombian creole steers.

Digestibility and intake are parameters difficult and expensive to estimate under grazing conditions; therefore, the aim of this study was to develop near-infrared reflectance spectroscopy (NIRS) calibrations applied to feces (F-NIRS) and evaluate their accuracy to predict dry matter digestibility (DMD) and dry matter intake (DMI) of Colombian creole cattle. Five digestibility trials using creole steers were conducted; indigestible neutral detergent fiber (iNDF) was used as internal marker and Cr2O3 and TiO2 as external markers. A total of 249 forage and 396 fecal samples from individual animals were collected, dried, and grinded for conventional chemical analysis. For spectral analysis, fecal samples were pooled across collection periods (77 samples). Chemometric analysis was performed using WinISI V4.10 software applying the modified partial least squares method. Cross-validation was performed to avoid overfitting the models. The goodness-of-fit statistics considered were the coefficient of determination in cross-validation and prediction sets (R2cv and r2, respectively) and the ratio performance deviation (RPD). Fecal NIRS calibrations developed for forage and supplement DMD showed a satisfactory fit (R2cv =0.87 and RPD=2.77 and R2cv=0.92 and RPD=3.50, respectively). The accuracy of fecal output equations using chromium (Cr) and titanium (Ti) was similar in terms of R2cv (0.92) and RPD (3.63 vs. 3.57). Total DMI equations using Ti performed better compared to Cr (R2cv = 0.82 vs. 0.78; RPD=2.41 vs. 2.17, respectively). The F-NIRS models were validated using a completely independent set of fecal samples showing a moderate fit (r2>0.8 and RPD>2.0). This study showed that F-NIRS is a feasible tool to predict DMD and DMI of creole steers under grazing conditions. However, previous to socialization, this requires an improvement in accuracy of the calibrated equations related to grazing animals in different production contexts.

Improved estimation of particulate matter in China based on multisource data fusion

Particulate matter (PM) is a global health concern and causes millions of premature deaths worldwide annually. High-resolution and full-coverage PM datasets are essential to support the accurate assessment of PM exposure. Here, a three-stage model framework is developed based on the Community Multiscale Air Quality (CMAQ) simulations (12 km) and multisource data fusion to estimate 1 km daily PM concentrations across China in 2015, including PM2.5 (<2.5 μm) and PM10 (<10 μm). The three-stage model performs well with cross-validation coefficient of determination (R2) of 0.91 and 0.87, and root mean square error (RMSE) of 17.3 μg/m3 and 27.2 μg/m3 for PM2.5 and PM10, respectively. After data fusion from multiple sources, the concentrations of PM2.5 and PM10 are in better agreement with ground observations compared to the CMAQ simulation with RMSE reduced by 72 % and 67 %. High PM2.5 events mainly occur in the North China Plain, Yangtze River Delta, and Sichuan Basin, and PM10 show similar spatial patterns to PM2.5 in eastern China. These full-coverage PM datasets enable in-depth analysis of PM pollution over small areas and support future epidemiological studies and health assessments.

Monitoring Compositional Changes in Black Soldier Fly Larvae (BSFL) Sourced from Different Waste Stream Diets Using Attenuated Total Reflectance Mid Infrared Spectroscopy and Chemometrics.

Black soldier fly (Hermetia illucens, L.) larvae are characterized by their ability to convert a variety of organic matter from food waste into a sustainable source of food (e.g., protein). This study aimed to evaluate the use of attenuated total reflectance (ATR) mid-infrared (MIR) spectroscopy to monitor changes in the composition as well as to classify black soldier fly larvae (BSFL) samples collected from two growth stages (fifth and sixth instar) and two waste stream diets (bread and vegetables, soy waste). The BSFL samples were fed on either a soy or bread-vegetable mix waste in a control environment (temperature 25 °C, and humidity 70%). The frass and BSFL samples harvested as fifth and sixth instar samples were analyzed using an ATR-MIR instrument where frequencies at specific wavenumbers were compared and evaluated using different chemometric techniques. The PLS regression models yield a coefficient of determination in cross-validation (R2) > 0.80 for the prediction of the type of waste used as diet. The results of this study also indicated that the ratio between the absorbances corresponding to the amide group (1635 cm−1) and lipids (2921 + 2849 cm−1) region was higher in diets containing a high proportion of carbohydrates (e.g., bread-vegetable mix) compared with the soy waste diet. This study demonstrated the ability of MIR spectroscopy to classify BSFL instar samples according to the type of waste stream used as a diet. Overall, ATR-MIR spectroscopy has shown potential to be used as tool to evaluate and monitor the development and growth of BSFL. The utilization of MIR spectroscopy will allow for the development of traceability systems for BSFL. These tools will aid in risk evaluation and the identification of hazards associated with the process, thereby assisting in improving the safety and quality of BSFL intended to be used by the animal feed industry.

An Intelligent Near-Infrared Diffuse Reflectance Spectroscopy Scheme for the Non-Destructive Testing of the Sugar Content in Cherry Tomato Fruit

As a new non-destructive testing technology, near-infrared spectroscopy has broad application prospects in agriculture, food, and other fields. In this paper, an intelligent near-infrared diffuse reflectance spectroscopy scheme (INIS) for the non-destructive testing of the sugar contents in vegetables and fruits was proposed. The cherry tomato were taken as the research object. The applicable objects and features of the three main methods of near-infrared detection were compared. According to the advantages and disadvantages of the three near infrared (NIR) detection methods, the experiment was carried out. This experiment involved the near-infrared diffuse reflection detection method, and the back propagation (BP) network model was established to research the sugar content of the cherry tomatoes. We used smoothing and a principal component analysis (PCA) to extract the final spectrum from the experimental spectrum. Taking the preprocessed spectral data as the input of the network and the measured sugar content of the cherry tomatoes as the output, the 80-12-1 network model structure was established. The cross-validation coefficient of determination was 0.8328 and the mean absolute deviation was 0.5711. The results indicate that the BP neural network can quickly and effectively detect the sugar content in cherry tomatoes. This intelligent near-infrared diffuse reflectance spectroscopy (INIS) scheme can be extended and optimized for almost all sugar-containing fruits and vegetables in the future.

Spatiotemporally Continuous Reconstruction of Retrieved PM2.5 Data Using an Autogeoi-Stacking Model in the Beijing-Tianjin-Hebei Region, China

Aerosol optical depth (AOD) observations have been widely used to generate wide-coverage PM2.5 retrievals due to the adverse effects of long-term exposure to PM2.5 and the sparsity and unevenness of monitoring sites. However, due to non-random missing and nighttime gaps in AOD products, obtaining spatiotemporally continuous hourly data with high accuracy has been a great challenge. Therefore, this study developed an automatic geo-intelligent stacking (autogeoi-stacking) model, which contained seven sub-models of machine learning and was stacked through a Catboost model. The autogeoi-stacking model used the automated feature engineering (autofeat) method to identify spatiotemporal characteristics of multi-source datasets and generate extra features through automatic non-linear changes of multiple original features. The 10-fold cross-validation (CV) evaluation was employed to evaluate the 24-hour and continuous ground-level PM2.5 estimations in the Beijing-Tianjin-Hebei (BTH) region during 2018. The results showed that the autogeoi-stacking model performed well in the study area with the coefficient of determination (R2) of 0.88, the root mean squared error (RMSE) of 17.38 µg/m3, and the mean absolute error (MAE) of 10.71 µg/m3. The estimated PM2.5 concentrations had an excellent performance during the day (8:00–18:00, local time) and night (19:00–07:00) (the cross-validation coefficient of determination (CV-R2): 0.90, 0.88), and captured hourly PM2.5 variations well, even in the severe ambient air pollution event. On the seasonal scale, the R2 values from high to low were winter, autumn, spring, and summer, respectively. Compared with the original stacking model, the improvement of R2 with the autofeat and hyperparameter optimization approaches was up to 5.33%. In addition, the annual mean values indicated that the southern areas, such as Shijiazhuang, Xingtai, and Handan, suffered higher PM2.5 concentrations. The northern regions (e.g., Zhangjiakou and Chengde) experienced low PM2.5. In summary, the proposed method in this paper performed well and could provide ideas for constructing geoi-features and spatiotemporally continuous inversion products of PM2.5.

Establishment of Non-Destructive Methods for the Detection of Amylose and Fat Content in Single Rice Kernels Using Near-Infrared Spectroscopy

For the efficient selection of high-quality rice varieties, the near-infrared spectroscopy (NIRS) technique has been widely applied to detect constituents in single rice kernels. Compared with other constituents, amylose content (AC) and fat content (FC) are the key parameters that can affect the quality of rice. Based on two modified AC and FC trace detection methods, two NIRS methods to detect AC and FC in single rice kernels were developed. Using the proposed methods, the AC and FC in two groups of rice kernel datasets were measured. The datasets were collected on two spectrometers with different sample movement states (static and dynamic) and measurement modes (diffuse reflectance (NIRr) and diffuse transmission (NIRt)). By optimizing the pre-treatment method and spectral range, the determination coefficients of cross-validation (R2cv) and prediction (R2p) of the NIRS models under different measurement conditions were all above 0.6. The results indicated that the proposed methods were applicable to the rapid, non-destructive detection and sorting of individual rice seeds with different AC and FC, and it was shown that these methods can meet the requirements of the rough screening of rice seed varieties.

Effect of pasteurization on coagulation properties of bovine milk and the role of major composition traits and protein fractions

Milk coagulation properties (MCP) worsens after heat treatment, however the specific mechanisms responsible have been scarcely explored. In this study, 100 milk samples were available to i) identify the raw milk characteristics responsible for unfavorable changes in MCP after pasteurization and ii) develop infrared prediction models for pasteurized milk MCP using spectra of raw samples. The loss in coagulation ability due to pasteurization was lower when raw milk had optimal MCP, higher acidity, greater protein content and lower β-lactoglobulin content. For the four MCP, the trait measured before pasteurization (raw milk) was the most important variable influencing the corresponding trait after heating. For example, rennet coagulation time (RCT), κ-casein, protein, lactose and pH of raw milk significantly affected pasteurized milk RCT (P < 0.001). For curd firmness, each unit (mm) corresponded to 58.65 g/100 g κ-casein. In general, raw milk β-lactoglobulin unfavorably affected pasteurized milk MCP (e.g., the estimate of curd firming time was 81.39 g/100 g). Results suggested that only the prediction model of RCT (pasteurized milk) achieved an exploitable coefficient of determination in cross-validation (0.66). Our outcomes are relevant for dairy plants manufacturing cheese from pasteurized standardized milk and could support producers’ decision-making.

Machine Learning Quantitative Structure-Property Relationships as a Function of Ionic Liquid Cations for the Gas-Ionic Liquid Partition Coefficient of Hydrocarbons.

Ionic liquids (ILs) are known for their unique characteristics as solvents and electrolytes. Therefore, new ILs are being developed and adapted as innovative chemical environments for different applications in which their properties need to be understood on a molecular level. Computational data-driven methods provide means for understanding of properties at molecular level, and quantitative structure–property relationships (QSPRs) provide the framework for this. This framework is commonly used to study the properties of molecules in ILs as an environment. The opposite situation where the property is considered as a function of the ionic liquid does not exist. The aim of the present study was to supplement this perspective with new knowledge and to develop QSPRs that would allow the understanding of molecular interactions in ionic liquids based on the structure of the cationic moiety. A wide range of applications in electrochemistry, separation and extraction chemistry depends on the partitioning of solutes between the ionic liquid and the surrounding environment that is characterized by the gas-ionic liquid partition coefficient. To model this property as a function of the structure of a cationic counterpart, a series of ionic liquids was selected with a common bis-(trifluoromethylsulfonyl)-imide anion, [Tf2N]−, for benzene, hexane and cyclohexane. MLR, SVR and GPR machine learning approaches were used to derive data-driven models and their performance was compared. The cross-validation coefficients of determination in the range 0.71–0.93 along with other performance statistics indicated a strong accuracy of models for all data series and machine learning methods. The analysis and interpretation of descriptors revealed that generally higher lipophilicity and dispersion interaction capability, and lower polarity in the cations induces a higher partition coefficient for benzene, hexane, cyclohexane and hydrocarbons in general. The applicability domain analysis of models concluded that there were no highly influential outliers and the models are applicable to a wide selection of cation families with variable size, polarity and aliphatic or aromatic nature.

Shedding light on human tissue (in vivo) to predict satiation, satiety, and food intake using near infrared reflectance spectroscopy: A preliminary study

The aim of this study was to evaluate the ability of a portable near infrared (NIR) instrument to collect the spectra in vivo of different tissues in healthy individuals and to relate their spectral information with food and energy intake, satiation, and satiety data. In this study, a hand-held NIR instrument was used to collect the spectra of different human tissues (e.g. arm, ear, face, jaw and wrist) with partial least squares (PLS) regression used to relate the NIR data with food and energy intake, satiation, and satiety measured in healthy individuals. The coefficient of determination in cross-validation (R2CV) and the standard error in cross validation (SECV) for the prediction of satiety ranged between 0.58 and 0.62 and 223.3–235.0 total area under the curve (AUC), respectively, depending on the tissue analysed. The PLS cross-validation models based on the NIR spectra collected in both the arm and face tissues gave the best prediction of food intake (R2CV 0.47–0.51, SECV 110.8-115 g). No workable calibrations were developed for the prediction of satiation, which might be associated with the inherent complexity of this parameter as well as the experimental conditions used to collect the data (e.g. type of tissue analysed). These results demonstrated the potential ability of in vivo NIR spectroscopy to identify tissue differences associated with satiety and food intake in individuals. However, a wider variety of food types, diets, and human subjects (samples) are needed to develop robust relationships between the NIR spectra of a tissue with both satiety and food intake.

Improving Spatial Disaggregation of Crop Yield by Incorporating Machine Learning with Multisource Data: A Case Study of Chinese Maize Yield

Spatially explicit crop yield datasets with continuous long-term series are essential for understanding the spatiotemporal variation of crop yield and the impact of climate change on it. There are several spatial disaggregation methods to generate gridded yield maps, but these either use an oversimplified approach with only a couple of ancillary data or an overly complex approach with limited flexibility and scalability. This study developed a spatial disaggregation method using improved spatial weights generated from machine learning. When applied to Chinese maize yield, extreme gradient boosting (XGB) derived the best prediction results, with a cross-validation coefficient of determination (R2) of 0.81 at the municipal level. The disaggregated yield at 1 km grids could explain 54% of the variance of the county-level statistical yield, which is superior to the existing gridded maize yield dataset in China. At the site level, the disaggregated yields also showed much better agreement with observations than the existing gridded maize yield dataset. This lightweight method is promising for generating spatially explicit crop yield datasets with finer resolution and higher accuracy, and for providing necessary information for maize production risk assessment in China under climate change.

Towards Inline Prediction of Color Development for Wood Stained with Chemical Stains Using Near-Infrared Spectroscopy

Abstract The chemical composition of wood determines the color development when applying chemical stains to the surface of wood. However, different species and individuals from the same species can show variations in the chemical composition, resulting in the risk of nonuniform color development in industrial staining processes between different batches of wood. In the present study, near-infrared (NIR) models were developed to predict wood specimen color development after applying three different concentrations of the chemical stains iron acetate and sodium bicarbonate. The modeling dataset included the NIR spectra of the untreated wood, stain treatment, concentration, and the International Commission on Illumination (CIE) L*a*b* color value before stain application for 210 specimens from five commercial wood species, including red oak (Quercus rubra), white oak (Quercus alba), yellow poplar (Liriodendron tulipifera), southern yellow pine (Pinus spp.), and western red cedar (Thuja plicata). The models were developed by partial least squares regression (PLSR), using 13 different mathematical transformations on the NIR spectra as well as the raw spectral data. Models with single stains and global-species/stain models were developed and compared. The models for iron acetate showed promising results in predicting the color development with the coefficient of determination for cross-validation ( ≥ 0.92), while the models for sodium bicarbonate showed acceptable results with of 0.71 to 0.89. However, a global model including both stains resulted in an unsatisfying prediction of the CIE L*a*b* color values, with of 0.46 to 0.76. The NIR models can be useful for online predictions of color development in industrial staining processes of wood with chemical stains.