Quantitative Determination of the Components in Corn and Tobacco Samples by Using Near-Infrared Spectroscopy and Multiblock Partial Least Squares

Multiblock partial least squares (MB-PLS) are applied for determination of corn and tobacco samples by using near-infrared diffuse reflection spectroscopy. In the model, the spectra are separated into several sub-blocks along the wavenumber, and different latent variable number was used for each sub-block. Compared with ordinary PLS, the importance and the contribution of each sub-block can be balanced by super-weights and the usage of different latent variable numbers. Therefore, the prediction obtained by the MB-PLS model is superior to that of the ordinary PLS, especially for the large data sets of tobacco samples with a large number of variables.     

Regression models based on new local strategies for near infrared spectroscopic data

In this work, a comparative study of two novel algorithms to perform sample selection in local regression based on Partial Least Squares Regression (PLS) is presented. These methodologies were applied for Near Infrared Spectroscopy (NIRS) quantification of five major constituents in corn seeds and are compared and contrasted with global PLS calibrations. Validation results show a significant improvement in the prediction quality when local models implemented by the proposed algorithms are applied to large data bases.     

Global,local and unique decompositions in OnPLS for multiblock data analysis

OnPLS is an extension of O2PLS that decomposes a set of matrices, in either multiblock or path model analysis, such that each matrix consists of two parts: a globally joint part containing variation shared with all other connected matrices, and a part that contains locally joint and unique variation, i.e. variation that is shared with some, but not all, other connected matrices or that is unique in a single matrix.     

How to construct a multiple regression model for data with missing elements and outlying objects

The aim of this study is to show the usefulness of robust multiple regression techniques implemented in the expectation maximization framework in order to model successfully data containing missing elements and outlying objects. In particular, results from a comparative study of partial least squares and partial robust M-regression models implemented in the expectation maximization algorithm are presented. The performances of the proposed approaches are illustrated on simulated data with and without outliers, containing different percentages of missing elements and on a real data set. The obtained results suggest that the proposed methodology can be used for constructing satisfactory regression models in terms of their trimmed root mean squared errors.     

Maintenance of a calibration model for near infrared spectrometry by a combined principal component analysis-partial least squares approach

A novel strategy for building and maintaining calibration models has been developed for use when the future boundaries of the sample set are unknown or likely to change. Such a strategy could have an impact on the economics and time required to obtain and maintain a calibration model for routine analysis. The strategy is based on both principal component analysis (PCA) and partial least squares (PLS) multivariate techniques. The principal action of the strategy is to define how “similar” a new sample is to the samples currently defining the calibration dataset. This step is performed by residuals analysis, following PCA. If the new sample is considered to have a spectrum “similar” to previously available spectra, then the model is assumed able to predict the analyte concentration. Conversely, if the new sample is considered “dissimilar”, then there is new information in this sample, which is unknown to the calibration model and the new sample is added automatically to the calibration set in order to improve the model. The strategy has been applied to a real industrial dataset provided by BP Amoco Chemicals. The data consists of spectra of 102 sequential samples of a raw material. The strategy produced an accurate calibration model for both target components starting with only the first four samples, and required a further 17 reference measurements to maintain the model for the whole sampling sequence, which was over a 1-year period.     

A new method of comparing 2D-PAGE maps based on the computation of Zernike moments and multivariate statistical tools

The aim of this work was to obtain the correct classification of a set of two-dimensional polyacrylamide gel electrophoresis map images using the Zernike moments as discriminant variables. For each 2D-PAGE image, the Zernike moments were computed up to a maximum p order of 100. Partial least squares discriminant analysis with variable selection, based on a backward elimination algorithm, was applied to the moments calculated in order to select those that provided the lowest error in cross-validation. The new method was tested on four datasets: (1) samples belonging to neuroblastoma; (2) samples of human lymphoma; (3) samples from pancreatic cancer cells (two cell lines of control and drug-treated cancer cells); (4) samples from colon cancer cells (total lysates and nuclei treated or untreated with a histone deacetylase inhibitor). The results demonstrate that the Zernike moments can be successfully applied for fast classification purposes. The final aim is to build models that can be used to achieve rapid diagnosis of these illnesses.     

Determination of organic additives in mortars by near-IR spectroscopy. A novel approach to designing a sample set with high-variability components

Industrial mortars consist primarily of a mixture of cement and an aggregate plus a small amount of additives that are used to modify specific properties. Using too high or too low additive rates usually results in the loss of desirable properties in the end product. This entails carefully controlling the amounts of additives added to mortar in order to ensure correct dosing and/or adequate homogeneity in the final mixture. Near-IR (NIR) spectroscopy has proved effective for this purpose as it requires no sample pretreatment and affords expeditious analyses. The purpose of this work was to determine two organic additives (viz. Ad1 and Ad2) in mortars by using partial least squares regression multivariate calibration models constructed from NIR spectroscopic data. The additives are used to expedite setting and increase cohesion between particles in the mortar. In order to ensure that the sample set contained natural variability in the samples, we used a methodology based on experimental design to construct a representative set of samples. This novel design is based on a hexagonal antiprism that encompasses the concentration ranges spanned by the analytes and the variability inherent in each additive. The D-optimality criterion was used to obtain various combinations between Ad1 and Ad2 additive classes. The partial least squares calibration models thus constructed for each additive provided accurate predictions: the intercept and the slope of the plots of predicted values versus reference values for each additive were close to 0 and 1, respectively, and their confidence ranges included the respective value. The ensuing analytical methods were validated by using an external sample set.     

Use of the bootstrap and permutation methods for a more robust variable importance in the projection metric for partial least squares regression

Bio-pharmaceutical manufacturing is a multifaceted and complex process wherein the manufacture of a single batch hundreds of processing variables and raw materials are monitored. In these processes, identifying the candidate variables responsible for any changes in process performance can prove to be extremely challenging. Within this context, partial least squares (PLS) has proven to be an important tool in helping determine the root cause for changes in biological performance, such as cellular growth or viral propagation. In spite of the positive impact PLS has had in helping understand bio-pharmaceutical process data, the high variability in measured response (Y) and predictor variables (X), and weak relationship between X and Y, has at times made root cause determination for process changes difficult. Our goal is to demonstrate how the use of bootstrapping, in conjunction with permutation tests, can provide avenues for improving the selection of variables responsible for manufacturing process changes via the variable importance in the projection (PLS-VIP) statistic. Although applied uniquely to the PLS-VIP in this article, the generality of the aforementioned methods can be used to improve other variable selection methods, in addition to increasing confidence around other estimates obtained from a PLS model.     

Quality assessment of gasoline using comprehensive two‐dimensional gas chromatography combined with unfolded partial least squares: A reliable approach for the detection of gasoline adulteration

Comprehensive two‐dimensional gas chromatography and flame ionization detection combined with unfolded‐partial least squares is proposed as a simple, fast and reliable method to assess the quality of gasoline and to detect its potential adulterants. The data for the calibration set are first baseline corrected using a two‐dimensional asymmetric least squares algorithm. The number of significant partial least squares components to build the model is determined using the minimum value of root‐mean square error of leave‐one out cross validation, which was 4. In this regard, blends of gasoline with kerosene, white spirit and paint thinner as frequently used adulterants are used to make calibration samples. Appropriate statistical parameters of regression coefficient of 0.996–0.998, root‐mean square error of prediction of 0.005–0.010 and relative error of prediction of 1.54–3.82% for the calibration set show the reliability of the developed method. In addition, the developed method is externally validated with three samples in validation set (with a relative error of prediction below 10.0%). Finally, to test the applicability of the proposed strategy for the analysis of real samples, five real gasoline samples collected from gas stations are used for this purpose and the gasoline proportions were in range of 70–85%. Also, the relative standard deviations were below 8.5% for different samples in the prediction set.     

Iterative weighting of multiblock data in the orthogonal partial least squares framework

The integration of multiple data sources has emerged as a pivotal aspect to assess complex systems comprehensively. This new paradigm requires the ability to separate common and redundant from specific and complementary information during the joint analysis of several data blocks. However, inherent problems encountered when analysing single tables are amplified with the generation of multiblock datasets. Finding the relationships between data layers of increasing complexity constitutes therefore a challenging task. In the present work, an algorithm is proposed for the supervised analysis of multiblock data structures. It associates the advantages of interpretability from the orthogonal partial least squares (OPLS) framework and the ability of common component and specific weights analysis (CCSWA) to weight each data table individually in order to grasp its specificities and handle efficiently the different sources of Y-orthogonal variation.     

Sum of ranking differences (SRD) to ensemble multivariate calibration model merits for tuning parameter selection and comparing calibration methods

Most multivariate calibration methods require selection of tuning parameters, such as partial least squares (PLS) or the Tikhonov regularization variant ridge regression (RR). Tuning parameter values determine the direction and magnitude of respective model vectors thereby setting the resultant predication abilities of the model vectors. Simultaneously, tuning parameter values establish the corresponding bias/variance and the underlying selectivity/sensitivity tradeoffs. Selection of the final tuning parameter is often accomplished through some form of cross-validation and the resultant root mean square error of cross-validation (RMSECV) values are evaluated. However, selection of a “good” tuning parameter with this one model evaluation merit is almost impossible. Including additional model merits assists tuning parameter selection to provide better balanced models as well as allowing for a reasonable comparison between calibration methods. Using multiple merits requires decisions to be made on how to combine and weight the merits into an information criterion. An abundance of options are possible. Presented in this paper is the sum of ranking differences (SRD) to ensemble a collection of model evaluation merits varying across tuning parameters. It is shown that the SRD consensus ranking of model tuning parameters allows automatic selection of the final model, or a collection of models if so desired. Essentially, the user’s preference for the degree of balance between bias and variance ultimately decides the merits used in SRD and hence, the tuning parameter values ranked lowest by SRD for automatic selection. The SRD process is also shown to allow simultaneous comparison of different calibration methods for a particular data set in conjunction with tuning parameter selection. Because SRD evaluates consistency across multiple merits, decisions on how to combine and weight merits are avoided. To demonstrate the utility of SRD, a near infrared spectral data set and a quantitative structure activity relationship (QSAR) data set are evaluated using PLS and RR.     

Voltammetric determination of dopamine in the presence of ascorbic and uric acids using partial least squares regression: determination of dopamine in human urine and plasma

A new differential pulse voltammetric method for dopamine determination at a bare glassy carbon electrode has been developed. Dopamine, ascorbic acid (AA) and uric acid (UA) usually coexist in physiological samples. Because AA and UA can be oxidized at potentials close to that of DA it is difficult to determine dopamine electrochemically, although resolution can be achieved using modified electrodes. Additionally, oxidized dopamine mediates AA oxidation and the electrode surface can be easily fouled by the AA oxidation product. In this work a chemometrics strategy, partial least squares (PLS) regression, has been applied to determine dopamine in the presence of AA and UA without electrode modification. The method is based on the electrooxidation of dopamine at a glassy carbon electrode in pH 7 phosphate buffer. The dopamine calibration curve was linear over the range of 1–313 μM and the limit of detection was 0.25 μM. The relative standard error (RSE %) was 5.28%. The method has been successfully applied to the measurement of dopamine in human plasma and urine.      

Chemometric data analysis of pollutants in wastewater—a case study

In this study, chemometric techniques such as cluster analysis (CA), discriminant analysis (DA), principal component analysis (PCA) and partial least squares (PLS) were used to analyse the wastewater dataset to identify the factors which affect the composition of sewage of domestic origin, spatial and temporal variations, similarity/dissimilarity among the wastewater characteristics of cis- and trans-drains and discriminating variables. Samples collected from 24 wastewater drains in Lucknow city and from three sites on Gomti river in the month of January/February, May, August and November during the period of 5 years (1994-1999) were characterized for 32 parameters. The multivariate techniques successfully described the similarities/dissimilarities among the sewage drains on the basis of their wastewater characteristics and sources signifying the effect of routine domestic/commercial activities in respective drainage areas. Spatial and seasonal variations in wastewater composition were also determined successfully. CA generated six groups of drains on the basis of similar wastewater characteristic. PCA provided information on seasonal influence and compositional differences in sewage generated by domestic and industrial waste dominated drains and showed that drains influenced by mixed industrial effluents have high organic pollution load. DA rendered six variables (TDS, alkalinity, F, TKN, Cd and Cr) discriminating between cis- and trans-drains. PLS-DA showed dominance of Cd, Cr, NO₃, PO₄ and F in cis-drains wastewater. The results suggest that biological-process based STPs could treat wastewater both from the cis- as well as trans-drains, however, prior removal of toxic metals will be required from the cis-drains sewage. Further, seasonal variations in wastewater composition and pollution load could be the guiding factor for determining the STPs design parameters. The information generated would be useful in selection of process type and in designing of the proposed sewage treatment plants (STPs) for safe disposal of wastewater.     

An assisted approach of the global optimization for the experimental conditions in capillary electrophoresis

Experimental conditions have effect on the separation of capillary electrophoresis (CE) directly. In this work, a set of index to describe the separation in CE was established properly. Based on a combination of genetic algorithm and least square support vector machine, an assisted approach of global optimization for experimental conditions was proposed for the first time, and it was applied to the separation of four synthetic compounds by CE in nonaqueous system. Under the optimum conditions obtained by this approach, the result of the experiment was satisfactory and proved that this novel approach was effective. Furthermore, we investigated the most important conditions that mainly affect the separation effectiveness of CE by partial least squares regression analysis. Because of the generalization of this new approach proposed, it can be applied to the optimization of other experimental processes.     

Genetic algorithm-based method for selecting conditions in multivariate determination of povidone-iodine using hand scanner

Application of hand scanner in multivariate quantification of povidone-iodine (PVI), as a popular antiseptic agent, in some of pharmaceutical products is presented. Brightness, contrast, and mixed gamma were the adjustable scanner parameters. For selection of optimum values of the scanner parameters, partial least squares (PLS) and multiple linear regression (MLR), coupled with genetic algorithm, were performed. For the selected variables, both MLR and PLS performances were similar and appropriate. From the results obtained, it was concluded that the simpler method of MLR could be successfully applied instead of PLS, which requires more statistical experience. The considered concentration range for PVI in the calibration and prediction samples was 0.0-10.0% (w/v). For the analysis of pharmaceutical samples, generalized standard addition method (GSAM) was applied (on the variables selected by GA) and desirable results were obtained. Relative standard error (RSE) of less than 8% was obtained for the majority of samples analyzed.     

Recognition of active ingredients in tablets by chemometric processing of X-ray diffractometric data

The paper presents an approach to use Partial Least Squares Discriminant Analysis (PLS-DA) on X-ray powder diffractometry (XRPD) dataset to build a model which recognizes a presence (or absence) of particular drug substance (acetaminophen) in unknown mixture (OTC tablet). The dataset consisted of 33 XRPD signals, measured for 12 pure substances and 21 tablets containing them in different quantitative and qualitative ratios, along with unknown excipients. The model was built with an external validation dataset chosen by Kennard-Stone algorithm. The RMSECV value was equal to 0.3461 (87.8% of explained variance) and external predictive error (RMSEP) was equal to 0.3123 (86.2% of explained variance). The result suggests that small but properly prepared training datasets give ability to construct well-working discriminant models on XRPD signals.     

Gas-chromatographic fatty-acid fingerprints and partial least squares modeling as a basis for the simultaneous determination of edible oil mixtures

Partial least squares modeling and gas-chromatographic fatty-acid fingerprints are reported as a method for the simultaneous determination of cottonseed, olive, soybean and sunflower edible oil mixtures. In this work, two sets of three- and four-component combinations of oils were prepared, hydrolyzed and the obtained free fatty acids analyzed by gas chromatography (GC) without any further derivatization. The normalized percentages of the myristic (14:0), palmitic (16:0), palmitoleic (16:1), stearic (18:0), oleic (18:1), linoleic (18:2) and linolenic (18:3) acids were chromatographically measured in samples and used for constructing calibration matrix. The cross-validation method was used to select the number of factors and the proposed methods were validated by using two sets of synthetic oil mixture samples. The relative standard error for each oil in mixture samples was less than 10%. This approach allows determining possible adulteration in each of the four edible oils.     

A useful approach for the differentiation of wines according to geographical origin based on global volatile patterns

In this study, the feasibility of solid‐phase extraction combined with gas chromatography and mass spectrometry in tandem with partial least squares discriminant analysis was evaluated as a useful strategy to differentiate wines according to geographical origin (Azores, Canary and Madeira Islands) and types (white, red and fortified wine) based on their global volatile patterns. For this purpose, 34 monovarietal wines from these three wine grape‐growing regions were investigated, combining the high throughput extraction efficiency of the solid‐phase extraction procedure with the separation and identification ability. The partial least squares discriminant analysis results suggested that Madeira wines could be clearly discriminated from Azores and Canary wines. Madeira wines are mainly characterized by 2‐ethylhexan‐1‐ol, 3,5,5‐trimethylhexan‐1‐ol, ethyl 2‐methylbutanoate, ethyl dl ‐2‐hydroxycaproate, decanoic acid, 3‐methylbutanoic acid, and (E)‐whiskey lactone, whereas 3‐ethoxypropan‐1‐ol, 1‐octen‐3‐ol, (Z)‐3‐hexenyl butanoate, 4‐(methylthio)‐1‐butanol, ethyl 3‐hydroxybutanoate, isoamyl lactate, 4‐methylphenol, γ‐octalactone and 4‐(methylthio)‐1‐butanol, are mainly associated with Azores and Canary wines. The data obtained in this study revealed that solid‐phase extraction combined with gas chromatography and quadrupole mass spectrometry data and partial least squares discriminant analysis provides a suitable tool to discriminate wines, both in terms of geographical origin as well as wine type and vintage.     

Near infrared spectroscopy as a rapid tool to measure volatile aroma compounds in Riesling wine: possibilities and limits

Volatile chemical compounds responsible for the aroma of wine are derived from a number of different biochemical and chemical pathways. These chemical compounds are formed during grape berry metabolism, crushing of the berries, fermentation processes (i.e. yeast and malolactic bacteria) and also from the ageing and storage of wine. Not surprisingly, there are a large number of chemical classes of compounds found in wine which are present at varying concentrations (ng L⁻¹ to mg L⁻¹), exhibit differing potencies, and have a broad range of volatilities and boiling points. The aim of this work was to investigate the potential use of near infrared (NIR) spectroscopy combined with chemometrics as a rapid and low-cost technique to measure volatile compounds in Riesling wines. Samples of commercial Riesling wine were analyzed using an NIR instrument and volatile compounds by gas chromatography (GC) coupled with selected ion monitoring mass spectrometry. Correlation between the NIR and GC data were developed using partial least-squares (PLS) regression with full cross validation (leave one out). Coefficients of determination in cross validation (R ²) and the standard error in cross validation (SECV) were 0.74 (SECV: 313.6 μg L⁻¹) for esters, 0.90 (SECV: 20.9 μg L⁻¹) for monoterpenes and 0.80 (SECV: 1658 μg L⁻¹) for short-chain fatty acids. This study has shown that volatile chemical compounds present in wine can be measured by NIR spectroscopy. Further development with larger data sets will be required to test the predictive ability of the NIR calibration models developed.     

Improved analysis of multivariate data by variable stability scaling: application to NMR-based metabolic profiling

Variable scaling alters the covariance structure of data, affecting the outcome of multivariate analysis and calibration. Here we present a new method, variable stability (VAST) scaling, which weights each variable according to a metric of its stability. The beneficial effect of VAST scaling is demonstrated for a data set of ¹H NMR spectra of urine acquired as part of a metabonomic study into the effects of unilateral nephrectomy in an animal model. The application of VAST scaling improved the class distinction and predictive power of partial least squares discriminant analysis (PLS-DA) models. The effects of other data scaling and pre-processing methods, such as orthogonal signal correction (OSC), were also tested. VAST scaling produced the most robust models in terms of class prediction, outperforming OSC in this aspect. As a result the subtle, but consistent, metabolic perturbation caused by unilateral nephrectomy could be accurately characterised despite the presence of much greater biological differences caused by normal physiological variation. VAST scaling presents itself as an interpretable, robust and easily implemented data treatment for the enhancement of multivariate data analysis.