scholarly journals Improving Prediction of Peroxide Value of Edible Oils Using Regularized Regression Models

Molecules ◽  
2021 ◽  
Vol 26 (23) ◽  
pp. 7281
Author(s):  
William E. Gilbraith ◽  
J. Chance Carter ◽  
Kristl L. Adams ◽  
Karl S. Booksh ◽  
Joshua M. Ottaway
Keyword(s):  
Regression Model ◽  
Ridge Regression ◽  
Near Infrared ◽  
Edible Oils ◽  
Reference Method ◽  
Elastic Net ◽  
Spectral Information ◽  
Pls Regression ◽  
Spectral Window ◽  
Wide Range

We present four unique prediction techniques, combined with multiple data pre-processing methods, utilizing a wide range of both oil types and oil peroxide values (PV) as well as incorporating natural aging for peroxide creation. Samples were PV assayed using a standard starch titration method, AOCS Method Cd 8-53, and used as a verified reference method for PV determination. Near-infrared (NIR) spectra were collected from each sample in two unique optical pathlengths (OPLs), 2 and 24 mm, then fused into a third distinct set. All three sets were used in partial least squares (PLS) regression, ridge regression, LASSO regression, and elastic net regression model calculation. While no individual regression model was established as the best, global models for each regression type and pre-processing method show good agreement between all regression types when performed in their optimal scenarios. Furthermore, small spectral window size boxcar averaging shows prediction accuracy improvements for edible oil PVs. Best-performing models for each regression type are: PLS regression, 25 point boxcar window fused OPL spectral information RMSEP = 2.50; ridge regression, 5 point boxcar window, 24 mm OPL, RMSEP = 2.20; LASSO raw spectral information, 24 mm OPL, RMSEP = 1.80; and elastic net, 10 point boxcar window, 24 mm OPL, RMSEP = 1.91. The results show promising advancements in the development of a full global model for PV determination of edible oils.

A simple multiple linear regression model in near infrared spectroscopy for soluble solids content of pomegranate arils based on stability competitive adaptive re-weighted sampling

2021 ◽  
pp. 096703352098236
Author(s):  
Zhaoqiong Jiang ◽  
Yiping Du ◽  
Fangping Cheng ◽  
Feiyu Zhang ◽  
Wuye Yang ◽  
...  
Keyword(s):  
Variable Selection ◽  
Regression Model ◽  
Near Infrared ◽  
Nir Spectroscopy ◽  
Soluble Solids ◽  
Pls Regression ◽  
Variable Selection Method ◽  
Solids Content ◽  
Pomegranate Arils ◽  
Mlr Model

The objective of this study was to develop a multiple linear regression (MLR) model using near infrared (NIR) spectroscopy combined with chemometric techniques for soluble solids content (SSC) in pomegranate samples at different storage periods. A total of 135 NIR diffuse reflectance spectra with the wavelength range of 950-1650 nm were acquired from pomegranate arils. Based upon sampling error profile analysis (SEPA), outlier diagnosis was conducted to improve the stability of the model, and four outliers were removed. Several pretreatment and variable selection methods were compared using partial least squares (PLS) regression models. The overall results demonstrated that the pretreatment method of the first derivative (1D) was very effective and the variable selection method of stability competitive adaptive re-weighted sampling (SCARS) was powerful for extracting feature variables. The equilibrium performance of 1D-SCARS-PLS regression model for ten times was similar to 1D-PLS regression model, so that the advantage of wavelength selection was inconspicuous in PLS regression model. However, the number of variables selected by 1D-SCARS was less to 9, which was enough to establish a simple MLR model. The performance of MLR model for SSC of pomegranate arils based on 1D-SCARS was receivable with the root-mean-square error of calibration set (RMSEC) of 0.29% and prediction set (RMSEP) of 0.31%. This strategy combining variable selection method with MLR may have a broad prospect in the application of NIR spectroscopy due to its simplicity and robustness.

scholarly journals The diffuse interstellar band around 8620 Å

2020 ◽  
Vol 645 ◽  
pp. A14
Author(s):  
H. Zhao ◽  
M. Schultheis ◽  
A. Recio-Blanco ◽  
G. Kordopatis ◽  
P. de Laverny ◽  
...  
Keyword(s):  
Near Infrared ◽  
Galactic Center ◽  
Random Noise ◽  
Gaussian Function ◽  
Specific Model ◽  
Perfect Agreement ◽  
Spectral Window ◽  
Gaussian Profile ◽  
Synthetic Spectra ◽  
Wide Range

Context. Diffuse interstellar bands (DIBs) are interstellar absorption features that widely exist in the optical and near-infrared wavelength range. DIBs play an important role in the lifecycle of the interstellar medium and can also be used to trace the Galactic structure. Aims. We developed a set of procedures to automatically detect and measure the DIB around 8620 Å (the Gaia DIB) for a wide range of temperatures. The method was tested on ~5000 spectra from the Giraffe Inner Bulge Survey (GIBS) that has a spectral window similar to that of the Gaia–RVS spectra. Based on this sample, we studied the correlation between the equivalent width (EW) of the Gaia DIB and the interstellar reddening E(J − KS) toward the inner Galaxy, as well as the DIB intrinsic properties. Methods. Our procedure automatically checks and eliminates invalid cases, and then applies a specific local normalization. The DIB profile is fit with a Gaussian function. Specifically, the DIB feature is extracted from the spectra of late-type stars by subtracting the corresponding synthetic spectra. For early-type stars we applied a specific model based on the Gaussian process that needs no prior knowledge of the stellar parameters. In addition, we provide the errors contributed by the synthetic spectra and from the random noise. Results. After validation, we obtained 4194 reasonable fitting results from the GIBS database. An EW versus E(J − KS) relation is derived as E(J − KS) = 1.875 (±0.152) × EW − 0.011 (±0.048), according to E(B − V)∕EW = 2.721, which is highly consistent with previous results toward similar sightlines. After a correction based on the Vista Variables in the Via Lactea (VVV) database for both EW and reddening, the coefficient derived from individual GIBS fields, E(J − KS)∕EW = 1.884 ± 0.225, is also in perfect agreement with literature values. Based on a subsample of 1015 stars toward the Galactic center within − 3° < b < 3° and − 6° < l < 3°, we determined a rest-frame wavelength of the Gaia DIB as 8620.55 Å. Conclusions. The procedures for automatic detection and measurement of the Gaia DIB are successfully developed and have been applied to the GIBS spectra. A Gaussian profile is proved to be a proper and stable assumption for the Gaia DIB as no intrinsic asymmetry is found. A tight linearity of its correlation with the reddening is derived toward the inner Milky Way, which is consistent with previous results.

scholarly journals Age Grading An. Gambiae and An. Arabiensis Using Near Infrared Spectra and Artificial Neural Networks

2018 ◽  
Author(s):  
Masabho P. Milali ◽  
Maggy T. Sikulu-Lord ◽  
Samson S. Kiware ◽  
Floyd E. Dowell ◽  
George F. Corliss ◽  
...  
Keyword(s):  
Regression Model ◽  
Near Infrared ◽  
Mean Squared Error ◽  
Estimation Accuracy ◽  
Pls Regression ◽  
Binary Classifier ◽  
Ann Model ◽  
Artificial Neural ◽  
Binary Classifiers ◽  
Age Grading

Background Near infrared spectroscopy (NIRS) is currently complementing techniques to age-grade mosquitoes. NIRS classifies lab-reared and semi-field raised mosquitoes into < or ≥ 7 days old with an average accuracy of 80%, achieved by training a regression model using partial least squares (PLS) and interpreted as a binary classifier. Methods and findings We explore whether using an artificial neural network (ANN) analysis instead of PLS regression improves the current accuracy of NIRS models for age-grading malaria transmitting mosquitoes. We also explore if directly training a binary classifier instead of training a regression model and interpreting it as a binary classifier improves the accuracy. A total of 786 and 870 NIR spectra collected from laboratory reared An. gambiae and An. arabiensis, respectively, were used and pre-processed according to previously published protocols. Based on ten-fold Monte Carlo cross-validation, an ANN regression model scored root mean squared error (RMSE) of 1.6 ± 0.2 for An. gambiae and 2.8 ± 0.2 for An. arabiensis; whereas the PLS regression model scored RMSE of 3.7 ± 0.2 for An. gambiae, and 4.5 ± 0.1 for An. arabiensis. When we interpreted regression models as binary classifiers, the accuracy of the ANN regression model was 93.7 ± 1.0 % for An. gambiae, and 90.2 ± 1.7 % for An. arabiensis; while PLS regression model scored the accuracy of 83.9 ± 2.3% for An. gambiae, and 80.3 ± 2.1% for An. arabiensis. We also find that a directly trained binary classifier yields higher age estimation accuracy than a regression model interpreted as a binary classifier. A directly trained ANN binary classifier scored an accuracy of 99.4 ± 1.0 for An. gambiae, and 99.0 ± 0.6% for An. arabiensis; while a directly trained PLS binary classifier scored 93.6 ± 1.2% for An. gambiae, and 88.7 ± 1.1% for An. arabiensis. Conclusion Training both regression and binary classification age models using ANNs yields models with higher estimation accuracies than when the same age models are trained using PLS. Regardless of the model architecture, directly trained binary classifiers score higher accuracy on classifying age of mosquitoes than a regression model translated as binary classifier. Therefore, we recommend training models to estimate age of An. gambiae and An. arabiensis using ANN model architectures and direct training of binary classifier instead of training a regression model and interpret it as a binary classifier.

scholarly journals The prediction of forage maize digestibility by near infrared reflection spectroscopy

1994 ◽  
Vol 42 (2) ◽  
pp. 105-113
Author(s):  
J.L. De Boever ◽  
J. Van Waes ◽  
B.G. Cottyn ◽  
C.V. Boucque
Keyword(s):  
Near Infrared ◽  
Reference Method ◽  
Near Infrared Reflectance ◽  
Reference Analysis ◽  
Forage Maize ◽  
Wide Range ◽  
Cellulose Digestibility ◽  
Highly Correlated ◽  
Infrared Reflection Spectroscopy

The potential of near infrared reflectance spectroscopy (NIRS) to predict organic matter digestibility (OMD) of fresh forage maize was examined. Cellulose digestibility, corrected to in vivo level, served as a reference method. Calibration was based on 261 samples, varying in OMD from 68.0 to 80.3% and validation occurred in 58 samples (71.7-75.9% OMD). With a scanning IA-500 monochromator the best equation, based on the second derivative of the reflected energy at wavelengths 1620 and 1664 nm, had a standard error of prediction (SEP) of 0.65%. The repeatability of the prediction amounted to 0.49% and was smaller than that of the reference analysis. The best equation, developed for a simulation of an IA-450 filter-apparatus, had a SEP of 0.74%. Cross-validation on the calibration set showed the validity of the calibrations for a wide range of digestibility. NIRS-predicted OMD was highly correlated with the reference OMD, whereas calculated OMD, based on constant digestion coefficients for the ears and stalk + leaves, did not show any relationship.

Comparison of Near-Infrared, Infrared, and Raman Spectroscopy for the Analysis of Heavy Petroleum Products

2000 ◽  
Vol 54 (2) ◽  
pp. 239-245 ◽  
Author(s):  
Hoeil Chung ◽  
Min-Sik Ku
Keyword(s):  
Near Infrared ◽  
Signal To Noise Ratio ◽  
Reference Method ◽  
Nir Spectroscopy ◽  
Petroleum Products ◽  
American Petroleum Institute ◽  
Visible Range ◽  
Pls Regression ◽  
Absorption Bands ◽  
Heavy Petroleum

Near-infrared (NIR) spectroscopy has been successfully applied to the determination of API (American Petroleum Institute) gravity of atmospheric residue (AR), which is the heaviest fraction in crude oil. This fraction is completely dark and very viscous. Preliminary studies involving Raman and infrared (IR) spectroscopies were also evaluated along with NIR spectroscopy. The Raman spectrum of AR was completely dominated by strong fluorescence from polycyclic aromatic hydrocarbons, called asphaltenes. IR spectroscopy provided reasonable spectral features; however, its spectral reproducibility was poorer and noisier than that of NIR. Although absorption bands in the NIR region were broad and less characterized, NIR provided better spectral reproducibility with higher signal-to-noise ratio (which is one of the most important parameters in quantitative calibration in comparison to Raman and IR spectroscopies). Partial least-squares (PLS) regression was utilized to develop calibration models. NIR spectra of AR samples were broad, and baselines were varying due to the strong absorption in the visible range. However, the necessary information was successfully extracted and correlated to the reference API gravity with the use of PLS regression. API gravities in the prediction set were accurately predicted with an SEP (standard error of prediction) of 0.22. Additionally NIR showed approximately three times better repeatability compared to the ASTM reference method, which directly influences the process control performance.

Critical evaluation of spectral information of benchtop vs. portable near-infrared spectrometers: quantum chemistry and two-dimensional correlation spectroscopy for a better understanding of PLS regression models of the rosmarinic acid content in Rosmarini folium

The Analyst ◽  
2017 ◽  
Vol 142 (3) ◽  
pp. 455-464 ◽  
Author(s):  
Christian G. Kirchler ◽  
Cornelia K. Pezzei ◽  
Krzysztof B. Beć ◽  
Sophia Mayr ◽  
Mika Ishigaki ◽  
...  
Keyword(s):  
Quantum Chemistry ◽  
Acid Content ◽  
Rosmarinic Acid ◽  
Regression Models ◽  
Near Infrared ◽  
Critical Evaluation ◽  
Correlation Spectroscopy ◽  
Spectral Information ◽  
Pls Regression ◽  
Two Dimensional

We show the importance of monitoring the performances of available NIR-spectrometers in every analytical area.

scholarly journals Quantification of leghaemoglobin content in pea nodules based on near infrared hyperspectral imaging spectroscopy and chemometrics

2018 ◽  
Author(s):  
Damien Eylenbosch ◽  
Benjamin Dumont ◽  
Vincent Baeten ◽  
Bernard Bodson ◽  
Pierre Delaplace ◽  
...  
Keyword(s):  
Regression Model ◽  
Hyperspectral Imaging ◽  
Near Infrared ◽  
Multispectral Imaging ◽  
Imaging Spectroscopy ◽  
Reference Method ◽  
Multiple Linear Regression Model ◽  
Validation Dataset ◽  
Full Spectrum ◽  
Imaging Device

Leghaemoglobin content in nodules is closely related to the amount of nitrogen fixed by the legume–rhizobium symbiosis. It is, therefore, commonly measured in order to assess the effect of growth-promoting parameters such as fertilisation on the symbiotic nitrogen fixation efficiency of legumes. The cyanmethaemoglobin method is a reference method in leghaemoglobin content quantification, but this method is time-consuming, requires accurate and careful technical operations and uses cyanide, a toxic reagent. As a quicker, simpler and non-destructive alternative, a method based on near infrared (NIR) hyperspectral imaging was tested to quantify leghaemoglobin in dried nodules. Two approaches were evaluated: (i) the partial least squares (PLS) approach was applied to the full spectrum acquired with the hyperspectral device and (ii) the potential of multispectral imaging was also tested through the preselection of the most relevant wavelengths and the building of a multiple linear regression model. The PLS approach was tested on mean spectra acquired from samples containing several nodules and acquired separately from individual nodules. Peas (Pisum sativum L.) were cultivated in a greenhouse. The nodules were harvested on four different dates in order to obtain variations in leghaemoglobin content. The leghaemoglobin content measured with the cyanmethaemoglobin method in fresh nodules ranged between 1.4 and 4.2 mg leghaemoglobin g–1 fresh nodule. A PLS regression model was calibrated on leghaemoglobin content measured with the reference method and mean NIR spectra of dried nodules acquired with a hyperspectral imaging device. On a validation dataset, the PLS model predicted the leghaemoglobin content in nodule samples well (R2 = 0.90, root mean square error of prediction = 0.26). The multispectral approach showed similar performance. Applied to individual nodules, the PLS model highlighted a wide variability of leghaemoglobin content in nodules harvested from the same plant. These results show that NIR hyperspectral imaging could be used as a rapid and safe method to quantify leghaemoglobin in pea nodules.

scholarly journals Evaluating and improving corn nitrogen fertilizer recommendation tools across the U.S. Midwest

2018 ◽  
Author(s):  
◽  
Curtis Joel Ransom
Keyword(s):  
Regression Model ◽  
Linear Relationship ◽  
Fertilizer Application ◽  
Development Stage ◽  
Elastic Net ◽  
N Fertilizer ◽  
Available N ◽  
Weather Information ◽  
Fertilizer Recommendation ◽  
Wide Range

Determining which corn (Zea mays L.) nitrogen (N) recommendation tools best predict the economically optimal N rate (EONR) would be valuable for maximizing profits and minimizing environmental consequences. The objectives of this research were to evaluate the performance of publicly-available N fertilizer recommendation tools across a wide range of soil and weather environments for 1) prescribing EONR for planting and split N fertilizer applications, 2) improve understanding of the economic and environmental impact of these tools, 3) improve N recommendation tools by integrating soil and weather information, and 4) improve N recommendation tools by combing multiple tools. The evaluation was conducted on 49 N response trials that spanned eight states and three growing seasons. Soil and plant samples, weather, and management information were collected using standardized procedures to allow for a side-by-side comparison of tools. Tool N recommendations were for fertilizer applications either atplanting or an inseason applied at V9 corn development stage. Only 11of 31 tool recommendations were weakly related to EONR (P [less than or equal to] 0.10 and r[2] [less than or equal to] 0.24). These tools related to EONR resulted in only 21-47% of sites within [plus or minus]30 kg N ha-1 of EONR. When considering partial profit for these 11 tools the average profitability relative to EONR range from -$56 to -155 ha-1. An environmental assessment of these 11 tools found there was no difference found between tools, with environmental costs ranging from -$49 to 55 ha-1 relative to EONR. Using an elastic net regression model to incorporate soil and weather information helped to improve six N recommendation tools. This improvement resulted in a stronger linear relationship with EONR (r[2] [less than or equal to] 0.20 but [less than or equal to] 0.39; P less than 0.01) and resulted in [greater than or equal to] 35% but [less than or equal to] 55 % of the sites within [plus or minus] 30 kg N ha-1 of EONR. Using other ways to improve tools included combing two or three unique tools. The best results for an at-planting N fertilizer recommendation occurred when three at-planting N recommendation tools were combined with all interactions included in the elastic net regression model. This combined recommendation tool had an improved significant linear relationship with EONR (r[2] = 0.46; P less than 0.001) compared with the best tool evaluated alone (an increase in r2 of 0.27). The best combination of N recommendation tools for a split N fertilizer application occurred when using three tools with a decision tree (r[2] = 0.45; P less than 0.001) over the best tool evaluated alone (an increase in r[2] of 0.18). However, while improvements to these publicly-available tools were noteworthy, over half of the variation in EONR was still unexplained. This was not surprising since many other factors that impact soil-crop N dynamics are unconsidered, including factors that occur after a sidedress N application.

Asymptotic for LS estimators in the EV regression model for dependent errors

Filomat ◽  
2017 ◽  
Vol 31 (15) ◽  
pp. 4845-4856
Author(s):  
Konrad Furmańczyk
Keyword(s):  
Regression Model ◽  
Functional Dependence ◽  
Dependent Data ◽  
Errors In Variables ◽  
Mixing Conditions ◽  
Asymptotic Results ◽  
Dependence Measure ◽  
Wide Range ◽  
Ev Regression Model ◽  
Linear And Nonlinear

We study consistency and asymptotic normality of LS estimators in the EV (errors in variables) regression model under weak dependent errors that involve a wide range of linear and nonlinear time series. In our investigations we use a functional dependence measure of Wu [16]. Our results without mixing conditions complete the known asymptotic results for independent and dependent data obtained by Miao et al. [7]-[10].

Sign in / Sign up

Export Citation Format

Share Document