Beyond Prediction: Extracting Relevant Information from near Infrared Spectra

1996 ◽  
Vol 4 (1) ◽  
pp. 75-84 ◽  
Author(s):  
P. Robert ◽  
M.-F. Devaux ◽  
D. Bertrand
Keyword(s):  
Near Infrared ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Relevant Information ◽  
Correlated Data ◽  
Spectral Interpretation ◽  
Highly Correlated ◽  
Prediction Techniques ◽  
Spectral Patterns

With the increase of near infrared (NIR) applications, numerous chemometric methods have been developed. Among the mathematical treatments available, principal comoponent analysis (PCA) is certainly the most well-known when considering highly correlated data. In the field of near infrared spectroscopy, it allows the study of spectra without deleting wavelengths and without making any preliminary assumptions on the data. One advantage of PCA lies in the graphical displays obtained and, more precisely, on the similarity maps and spectral patterns. While the maps reveal clusters of the samples, the spectral patterns make a spectral interpretation possible. The present paper reviews our contribution to the development and application of PCA to NIR spectroscopy. It shows that PCA is the core of various mathematical treatments such as principal component regression (PCR), factorial discriminant analysis (FDA) and canonical correlation analysis (CCA). One advantage of using PCA in the prediction techniques lies in the use of all the wavelengths in the predictive model. The extraction of relevant and comprehensive wavelengths can be guided by CCA which allows the description of the samples by taking both mid- and near infrared data into account. Besides a comprehensive presentation of the mathematical treatements, examples are given.

scholarly journals Separation of Cultivars of Soybeans by Chemometric Methods Using Near Infrared Spectroscopy

2018 ◽  
Vol 10 (4) ◽  
pp. 351
Author(s):  
João S. Panero ◽  
Henrique E. B. da Silva ◽  
Pedro S. Panero ◽  
Oscar J. Smiderle ◽  
Francisco S. Panero ◽  
...  
Keyword(s):  
Spectral Data ◽  
Near Infrared ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Hierarchical Cluster ◽  
Relevant Information ◽  
Whole Grains ◽  
Signal Correction ◽  
Different Types ◽  
Linear Fit

Near Infrared (NIR) Spectroscopy technique combined with chemometrics methods were used to group and identify samples of different soy cultivars. Spectral data, collected in the range of 714 to 2500 nm (14000 to 4000 cm-1), were obtained from whole grains of four different soybean cultivars and were submitted to different types of pre-treatments. Chemometrics algorithms were applied to extract relevant information from the spectral data, to remove the anomalous samples and to group the samples. The best results were obtained considering the spectral range from 1900.6 to 2187.7 nm (5261.4 cm-1 to 4570.9 cm-1) and with spectral treatment using Multiplicative Signal Correction (MSC) + Baseline Correct (linear fit), what made it possible to the exploratory techniques Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) to separate the cultivars. Thus, the results demonstrate that NIR spectroscopy allied with de chemometrics techniques can provide a rapid, nondestructive and reliable method to distinguish different cultivars of soybeans.

The Effect of Multiplicative Scatter Correction (MSC) and Linearity Improvement in NIR Spectroscopy

1988 ◽  
Vol 42 (7) ◽  
pp. 1273-1284 ◽  
Author(s):  
Tomas Isaksson ◽  
Tormod Næs
Keyword(s):  
Near Infrared ◽  
Scatter Correction ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Data Sets ◽  
Prediction Errors ◽  
Regression Methods ◽  
Multiplicative Scatter Correction ◽  
Linearity Improvement

Near-infrared (NIR) reflectance spectra of five different food products were measured. The spectra were transformed by multiplicative scatter correction (MSC). Principal component regression (PCR) was performed, on both scatter-corrected and uncorrected spectra. Calibration and prediction were performed for four food constituents: protein, fat, water, and carbohydrates. All regressions gave lower prediction errors (7–68% improvement) by the use of MSC spectra than by the use of uncorrected absorbance spectra. One of these data sets was studied in more detail to clarify the effects of the MSC, by using PCR score, residual, and leverage plots. The improvement by using nonlinear regression methods is indicated.

Use of Near-Infrared Spectroscopy to Determine the Composition of High-Density/Low-Density Polyethylene Blend Films

1993 ◽  
Vol 47 (2) ◽  
pp. 222-228 ◽  
Author(s):  
Charles E. Miller
Keyword(s):  
Near Infrared ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
High Density ◽  
Low Density Polyethylene ◽  
Low Density ◽  
Blend Films ◽  
Polyethylene Blend

The ability of near-infrared (NIR) spectroscopy, combined with principal component regression (PCR), to nondestructively determine the blend ratio of high-density polyethylene (HDPE) and low-density polyethylene (LDPE) in extruded films is demonstrated. Results indicate that the NIR spectrum in the region 2100 to 2500 nm can be used to determine the HDPE mass percentage of 60–80- μm-thick film samples to within 2.5%, over a range of 0 to 100%. NIR spectral effects from scattering are important for the determination of the HDPE % for HDPE contents above 50%, and spectral effects from changes in the methyl group concentration and perhaps the PE crystallinity are important for the determination of the HDPE % for HDPE contents below 50%. In addition, a large variation between the spectra of replicate samples, probably caused by variations in the degree or direction of molecular orientation in the samples, was observed.

Nondestructive Characterizations of Polyethylene/Nylon Laminates by Near-Infrared Spectroscopy

1993 ◽  
Vol 47 (3) ◽  
pp. 346-356 ◽  
Author(s):  
Charles E. Miller ◽  
Svend A. Svendsen ◽  
Tormod Næ
Keyword(s):  
Layer Thickness ◽  
Food Packaging ◽  
Near Infrared ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Polyamide 6 ◽  
Chemical Modeling ◽  
Ethylene Vinyl Alcohol ◽  
Laminate Thickness

The use of near-infrared (NIR) spectroscopy for the rapid and nondestructive analysis of food packaging laminates containing polyethylene (PE), polyamide-6 (PA-6), and ethylene vinyl alcohol (EVOH) layers is demonstrated. The method of Pathlength Correction with Chemical Modeling (PLC-MC) is used to estimate the total laminate thickness, and Principal Component Regression (PCR), is used to estimate the thickness percentages of PE and EVOH in the laminates, from NIR reflective-transmission spectra in the region 1500–2500 nm. Results indicate that the NIR method can be used to determine the total laminate thickness within 2–4 μm, the PE layer thickness percentage within 0.7–1.8%, and the EVOH layer thickness within 0.7–0.8 μm. In addition, detailed observation of the PCR models indicates that the NIR method is also sensitive to the absorbed water content, the morphology of the polymers, and perhaps the amount of polyurethane adhesive in the laminates. The usefulness of PCR outlier detection, for identification and characterization of strange samples, and principal component rotation, for improvement of PCR model interpretability, is also demonstrated.

Comparison of Multivariate Calibration and Discriminant Analysis in Evaluating NIR Spectroscopy for Determination of Meat Tenderness

1997 ◽  
Vol 51 (3) ◽  
pp. 350-357 ◽  
Author(s):  
Tormod Næs ◽  
Kjell Ivar Hildrum
Keyword(s):  
Discriminant Analysis ◽  
Reference Values ◽  
Near Infrared ◽  
Multivariate Calibration ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Meat Tenderness ◽  
Intermediate Group ◽  
Calibration Methods

Often the primary goal of analytical measurement tasks is not to find good estimates of continuous reference values but rather to determine whether a sample belongs to one of a number of categories or subgroups. In this paper the potential of different statistical techniques in the classification of raw beef samples in tenderness subgroups was studied. The reference values were based on sensory analysis of beef tenderness of 90 samples from bovine M. longissimus dorsi muscles. The sample set was divided into three categories—very tough, intermediate, and very tender—according to degree of tenderness. A training set of samples was used to find the relationship between category and near-infrared (NIR) spectroscopic measurements. The study indicates that classical discriminant analysis has advantages in comparison to multivariate calibration methods [i.e., principal component regression (PCR)], in this application. One reason for this observation seems to be that PCR underestimates high measurement values and overestimates low values. In this way most samples are assigned to the intermediate group of samples, causing a small number of erroneous classifications for the intermediate subgroup, but a large number of errors for the two extreme groups. With the use of PCR the number of correct classifications in the extreme subgroups was as low as 23%, while the use of discriminate analysis increased this number to almost 60%. The number of classifications in correct or neighbor subgroup for the two extreme subgroups was equal to 97%. A “bias-correction” was also attempted for PCR, and this gave results comparable to the best results obtained by discriminant analysis methods. Test sets used NIR analysis of fresh, raw beef samples with different processing. While this spectroscopic approach had previously been shown to be useful with frozen products, it appears unsuitable at this time for fresh beef. However, its marginal analytical utility proved useful in evaluating the two classification approaches employed in this study.

Rapid identification of plant- and chemical-dyed cotton fabrics using the near-infrared technique

2020 ◽  
Vol 90 (19-20) ◽  
pp. 2275-2283
Author(s):  
Mingxia Li ◽  
Guangting Han ◽  
Wei Jiang ◽  
Chengfeng Zhou ◽  
Yuanming Zhang ◽  
...  
Keyword(s):  
Near Infrared ◽  
Recognition Rate ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Rapid Identification ◽  
Cotton Fabrics ◽  
Calibration Model ◽  
Pattern Recognition Methods ◽  
Validation Set

Plant dye is a promising dyestuff to be used in textiles due to its unique environmental compatibility. However, currently there is no effective method for the identification of plant-dyed and chemical-dyed textiles. In this study, near-infrared (NIR) spectroscopy combined with three kinds of pattern recognition methods, namely soft independent modeling of class analogy (SIMCA), partial least squares (PLS) regression and principal component regression (PCR), were applied to identify cotton fabrics dyed with plant and chemical dyes. A total of 336 plant dye and chemical dye dyed cotton fabrics were prepared and the NIR spectra were collected; 267 samples were used as the calibration set, while the remaining 69 samples were used as the validation set. After pretreatment with the Savitzky–Golay first derivative, the calibration model was constructed. In the SIMCA model, the correct recognition rate values of the calibration and prediction sets were 100% and 98.55%, respectively. The PLS model showed that the number of principal components (PCs) and the correlation coefficient ( R2) were 8 and 0.9978, respectively, and the results of PCR were PC = 10, R2 = 0.9937. Both methods were satisfactory for the predicted results. The overall results indicated that NIR spectroscopy could be used for rapid and nondestructive identification of plant-dyed cotton fabrics and chemical-dyed cotton fabrics.

scholarly journals Evaluation of ultraviolet, visible, and near infrared spectroscopy for the analysis of wine compounds

2014 ◽  
Vol 32 (No. 1) ◽  
pp. 37-47 ◽  
Author(s):  
M.J. Martelo-Vidal ◽  
M. Vázquez
Keyword(s):  
Tartaric Acid ◽  
Malic Acid ◽  
Near Infrared ◽  
Uv Spectroscopy ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Pure Compounds ◽  
Modelling Strategies ◽  
Ethanol Determination

Spectroscopy of UV-VIS-NIR combined with chemometric analyses was used as a non-destructive technique to build models for the quantitative characterisation of the main compounds of wine. The work in mixtures can give insight into how interferences affect the performance of calibrations in wines. Ethanol, glycerol, glucose, tartaric acid, malic acid, lactic acid, and acetic acid were evaluated as pure compounds and in mixtures. Different pre-treatments for the spectra and modelling strategies such as partial least squares (PLS) regression or Principal Component Regression (PCR) were evaluated. All pure compounds studied showed a good relationship between spectra and concentrations. However, interferences were observed in the mixtures and only good models for ethanol, tartaric acid, and malic acid were obtained. The best model was obtained in the NIR region for ethanol and in the UV region for tartaric acid and malic acid. The results indicate that NIR spectroscopy could be used as an alternative to conventional chemical methods for ethanol determination and UV spectroscopy for the determination of tartaric acid and malic acid.

A Rapid and Nondestructive Method for the Prediction of Ibuprofen Tablet Hardness Using NIR Diffuse Reflectance Spectroscopy

2019 ◽  
Vol 15 (5) ◽  
pp. 439-446
Author(s):  
Qiushi Peng ◽  
Yi Bao ◽  
Tingyu Chen ◽  
Qianrong Peng ◽  
Min Yang
Keyword(s):  
Near Infrared ◽  
Diffuse Reflectance Spectroscopy ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Magnesium Stearate ◽  
Nondestructive Method ◽  
Pcr Analysis ◽  
Sodium Magnesium ◽  
Tablet Hardness

Introduction: This study aimed at developing a technology to measure the hardness of Ibuprofen (IBU) tablets and optimize the IBU formulation using Near-infrared (NIR) spectroscopy. Materials and Methods: Tablets (400 mg±5%, 10mm in diameter) consisting of IBU, microcrystalline cellulose SH-103, carboxymethyl starch sodium, magnesium stearate, silicon dioxide were formed of various hardness (2kg, 4kg, 6kg, 8kg, 10kg, 12kg). The reflectance NIR spectra of various tablets were employed to establish 9 calibrations models, which were further used to predict tablet hardness by Partial least squares (PLS) and principal component regression (PCR) analysis. Results and Conclusion: Cross-validation with independent samples shows that PLS is the optimal predictive model. Which R2=0.9832, RSECV=0.334 and RSE=0.0669. This study established a new, simple, rapid, nondestructive and reliable methodology to optimize the IBU tablet hardness.

Canonical Correlation Analysis of Mid- and Near-Infrared Oil Spectra

1993 ◽  
Vol 47 (7) ◽  
pp. 1024-1029 ◽  
Author(s):  
M. F. Devaux ◽  
P. Robert ◽  
A. Qannari ◽  
M. Safar ◽  
E. Vigneau
Keyword(s):  
Correlation Analysis ◽  
Canonical Correlation Analysis ◽  
Infrared Spectra ◽  
Canonical Correlation ◽  
Near Infrared ◽  
Principal Component ◽  
Canonical Variate ◽  
Mid Infrared ◽  
Highly Correlated ◽  
Spectral Patterns

A mathematical procedure based on Canonical Correlation Analysis (CCA) was used in order to assign the wavelengths of the near-infrared spectra through knowledge of the mid-infrared spectra. The relevance of the treatment was tested on commercial oils that mainly differ in their level of unsaturation. Initially, two separated Principal Component Analyses (PCAs) were performed on the near- and mid-infrared data to overcome the high intercorrelations across the wavelengths. CCA was then applied to the resulting principal components. Near- and mid-infrared canonical variates were assessed so that they achieved maximum correlation. The procedure makes it possible to draw CCA spectral patterns that exhibit significant positive and negative peaks. The first near-infrared canonical variate was highly correlated with the first mid-infrared canonical variate ( r2 = 0.97). The corresponding near- and mid-infrared CCA spectral patterns were therefore given the same interpretation. The mid-infrared pattern opposed negative peaks characteristic of CH2 groups to the positive peaks of CH3 and =CH groups. Consequently, in the near-infrared pattern, the positive peaks at 1708, 2140, 2170, and 2480 nm were assigned to CH3 or =CH groups, and the negative peaks at 2304, 2344, and 2445 nm were assigned to CH2 groups. A more precise interpretation was obtained by comparing the wavelengths observed to theoretical values and to previous assignments.

Significance of Temperature Control in FT-NIR Spectrometers

1993 ◽  
Vol 47 (1) ◽  
pp. 7-11 ◽  
Author(s):  
C. A. Young ◽  
K. Knutson ◽  
J. D. Miller
Keyword(s):  
Near Infrared ◽  
Process Analysis ◽  
Cooling Water ◽  
Principal Component Regression ◽  
Nir Spectroscopy ◽  
Principal Component ◽  
Temperature Fluctuations ◽  
Absorbing Materials ◽  
Purge Gas ◽  
Chemometric Algorithms

Recent applications of near-infrared (NIR) spectroscopy to process analysis based on principal component regression or other chemometric algorithms have shown that NIR spectrometers become unstable after extended use. In the present study, stability problems are reported to occur within a 24-hour period. The instability was found to be at a maximum when highly IR-absorbing materials were analyzed and was attributed to temperature fluctuations in both the cooling water to the source and the purge gas to the optical bench. The problem was alleviated by using a thermostated recirculator and liquid nitrogen boil-off. Ensuing discussions show the problem to be relevant to other instruments which perform Fourier transformations on interferograms.

Sign in / Sign up

Export Citation Format

Share Document