Partial Least Squares Regression Regression Research Articles

Discrepancies in methodologies to determine chill requirements in temperate fruit trees constrain guidelines for future plantings in a global warming context

Temperate fruit trees enter a dormant state and resume growth development once cultivar-specific chill requirements (CR) are fulfilled. Sustainable fruit production requires crucial knowledge of the CR of the varieties to predict the adaptability of crops to future climate conditions. Experimental methodologies based on exposing shoots to forcing conditions, and statistical methods based on analysing long-term phenology records, such as Partial Least Square regression (PLS), are commonly used to estimate CR. To demonstrate the importance of unifying methodologies, we compared CR estimations from six approaches that included bud weight increase (30 % and 50 %), different percentages of bud break (30 %, 50 % and 90 %) after forcing conditions, and PLS regression. The CR of four almond and four apricot cultivars were experimentally and statistically determined in Murcia, South-Eastern Spain. Comparisons between approaches revealed that the choice of the method had a significant effect on the estimates. Large variation in CR estimates was found between PLS regression and the experimental methodologies in some late-flowering cultivars. To assess the adaptation potential of a certain cultivar to future climate conditions based on the CR estimates obtained through the different methodologies, we compared the probability of fulfilling cultivar-specific CR's under two global warming scenarios. Projections revealed a marked decreasing trend of chill availability in the region, which is already threatening the cultivation of genotypes with high CR. This study demonstrates that the variability of methodologies used to quantify CR has an important impact on assessing the probability of a successful adaptation. CR obtained by a certain methodology suggest that insufficient chill might be an increasingly common situation for some cultivars, potentially resulting in production risks. Conversely, other methodologies appear to indicate a more favourable performance for the same cultivar. Implementing standardized methods and metrics is essential to ensure more reliable information regarding the prospects for future crop adaptation.

PLS Regression Coupled with FTIR Analysis as a Fast Tool to Assess Properties and Nutrient Pools of Biochar-Based Fertilizers

ABSTRACT Biochar-based fertilizers (BBFs) agronomic performance depends on its nutrient pools, electrical conductivity (EC) and pH, conventionally assessed by laboratory classical wet methods that are non-environmentally friendly, time-consuming, and costly. However, the nutrient pools and properties of BBFs can be predicted by partial least square regression (PLS) based on Fourier Transform Infrared Spectroscopy (FTIR) analysis. Thus, we aimed to use PLS regression based on FTIR analysis to predict the pH, EC and nitrogen (N), phosphorus (P), and potassium (K) pools in BBFs produced by different routes. Sixteen BBFs were produced by mixing, previously or after the pyrolysis process of low-grade phosphate acidulated with different inorganic acids (sulfuric, hydrochloric and nitric acids), addition or no different sources of Mg (magnesium chloride and serpentinite). PLS models based on the normalized FTIR dataset in BBFs are considered good when R2 during calibration and cross-validation phases are higher than 0.81, values founded in our study to prediction of pH, EC, total, and mineral N, total P, and P soluble in water, citric acid, formic acid, and neutral ammonium citrate, and total K of BBFs. Besides, the difference between R2 of calibration and cross-validation lower than 0.2 founded indicated no overfitting. Furthermore, the PLS models had high prediction capacity (R2 of prediction higher than 0.55) and high accuracy (R2 to y-randomization <0.50). Thus, PLS regression models based on the normalized FTIR dataset in BBFs were models characterized by the good performance, robustness, and nonrandom able to predict the nutrient pools, pH, and EC in BBFs.

Mechanisms of propeller jet-induced migration, release, and distribution of perfluoroalkyl acids in sediment–water systems

Perfluoroalkyl acids (PFAAs) are continuously accumulated in surface sediments due to extensive and long-term application. However, the mechanisms through which disturbances induced by ship propeller jets at the riverbed cause secondary release of PFAAs from sediments remain unclear. In this study, the effects of different propeller rotational speeds on PFAA migration, release, and distribution in multiphase media were investigated by performing indoor flume experiments combined with particle tracking velocimetry. Moreover, key factors influencing PFAA migration and distribution were identified, and partial least squares regression (PLS) method was applied to develop quantitative prediction models of relationships among hydrodynamics, physicochemical parameters, and PFAA distribution coefficients. The total PFAA concentrations (ΣPFAAs) in overlying water under propeller jet action exhibited transient characteristics and hysteresis with time after the disturbance. In contrast, the ΣPFAAs in suspended particulate matter (SPM) exhibited an upward trend throughout the process with consistent characteristics. The spatial distribution trends of PFAAs in overlying water and SPM at different propeller rotational speeds featured vertical variability and axial consistency. Furthermore, PFAA release from sediments was driven by axial flow velocity (Vx) and Reynolds normal stress Ryy, while PFAA release from porewater was inextricably linked to Reynolds stresses Rxx, Rxy, and Rzz (p < 0.05). PLS regression models showed that variations in Vorticity, dissolved organic carbon, and pH influenced the decreases in PFAA distribution coefficients between SPM and overlying water (KD-SW) as propeller rotational speed increased, except for very long-chain PFAAs (C > 10). The increases in PFAA distribution coefficients between sediment and porewater (KD-SP) were mainly determined by physicochemical parameters of sediments, and the direct effect of hydrodynamics was relatively weak. Our study provides valuable information regarding the migration and distribution of PFAAs in multiphase media under propeller jet disturbance (both during and after disturbance).

Estimation of Aboveground Biomass of Potatoes Based on Characteristic Variables Extracted from UAV Hyperspectral Imagery

Aboveground biomass (AGB) is an important indicator for crop-growth monitoring and yield prediction, and accurate monitoring of AGB is beneficial to agricultural fertilization management and optimization of planting patterns. Imaging spectrometer sensors mounted on unmanned aerial vehicle (UAV) remote-sensing platforms have become an important technical method for monitoring AGB because the method is convenient, rapidly collects data and provides image data with high spatial and spectral resolution. To confirm the feasibility of UAV hyperspectral remote-sensing technology to estimate AGB, this study acquired hyperspectral images and measured AGB data over the potato bud, tuber formation, tuber growth, and starch-storage periods. The canopy spectrum obtained in each growth period was smoothed by using the Savitzky–Golay filtering method, and the spectral-reflection feature parameters, spectral-location feature parameters, and vegetation indexes were extracted. First, a Pearson correlation analysis was performed between the three types of characteristic spectral parameters and AGB, and the spectral parameters that reached a significant level of 0.01 in each growth period were selected. Next, the spectral parameters reaching a significance of 0.01 were optimized and screened by moving window partial least squares (MWPLS), Monte Carlo uninformative variable elimination (MC-UVE), and random frog (RF) methods, and the final model parameters were determined according to the thresholds of the root mean square error of cross-validation (RMSEcv), the reliability index, and the selected probability. Finally, the three optimal characteristic spectral parameters and their combinations were used to estimate the potato AGB in each growth period by combining the partial least squares regression (PLSR) and Gaussian process regression (GPR) methods. The results show that, (i) ranked from high to low, vegetation indexes, spectral-location feature parameters, and spectral-reflection feature parameters in each growth period are correlated with the AGB, and these correlations all first improve and then degrade in going from the budding period to the starch-storage period. (ii) The AGB estimation model based on the characteristic variables screened by the three methods in each growth period is most accurate with RF, less so with MC-UVE, and least accurate with MWPLS. (iii) Estimating the AGB with the same variables combined with the PLSR method in each growth period is more accurate than the corresponding GPR method, but the estimations produced by the two methods both show a trend of first improving and then worsening from the budding period to the starch-accumulation period. The accuracy of the estimation models constructed by PLSR and GPR from high to low is based on comprehensive variables, vegetation indexes, spectral-location feature parameters and spectral-reflection feature parameters. (iv) When combined with the RF-PLSR method to estimate AGB in each growth period, the best R2 values are 0.65, 0.68, 0.72, and 0.67, the corresponding RMSE values are 167.76, 162.98, 160.77, and 169.24 kg/hm2, and the corresponding NRMSE values are 19.76%, 16.01%, 15.04%, and 16.84%. The results of this study show that a variety of characteristic spectral parameters may be extracted from UAV hyperspectral images, that the RF method may be used for optimizing and screening, and that PLSR regression provides accurate estimates of the potato AGB. The proposed approach thus provides a rapid, accurate, and nondestructive way to monitor the growth status of potatoes.

Preferential sorption of polysaccharides on mackinawite: A chemometrics approach

Marine sediments represent the most important sink for natural organic matter (NOM) across geological time spans. Strong associations between minerals, including iron oxides, and organic matter reaching the seafloor play a fundamental role in this preservation and have been known for some decades. Despite the importance of this protective mechanism in the balances of the global carbon budget, very little is known about the affinity of NOM for reduced iron species such as mackinawite (FeS) in the anoxic layers of sediment. In this study, equilibrium partition coefficients (Kd) for three types of NOM (soil extract, corn leaf extract and plankton extract) on FeS were determined through batch sorption experiments. Models were then built via multiple linear regression (MLR) and partial least squares regression (PLSR) in order to predict sorption Kd’s based on the chemical characteristics of the NOM. Investigation of the PLSR regression coefficients indicates that functional groups characteristic of polysaccharides are the greatest positive predictors of NOM sorption onto FeS at marine sediment porewater pH. This conclusion was independently verified by analyses of NOM FTIR spectra pre- and post-sorption. PLSR outperformed MLR with a lower root mean square error of prediction (RMSEP) of 53.4 L/kg compared to 64.0 L/kg. To our knowledge, this research presents a novel machine-learning approach to the quantitative modelling of NOM sorption to minerals found in anoxic marine environments.

PLS CALCULATION OF FTIR AND SPECTROPHOTOMETRIC METHODS FOR DETERMINATION SIMULTANEOUS OF DEXTROMETHORPHAN HBR AND GLYCERYL GUAIACOLATE IN TABLET MIXTURE

Practical, easy, correct, and perfection of Partial least squares regression (PLS) calculations with Fourier transform infrared (FTIR) and Spectrophotometric methods to the simultaneous determination of Dextromethorphan HBr and Glyceryl Guaiacolate mixtures without separation in tablet formulations. The research was using PLS regression for FTIR at a wavelength of 806.24-844.82 nm and for dextromethorphan HBr and on 705.94-744.52 nm for Glyceryl Guaiacolate, while by spectrophotometric method at 230-300 nm with 2 nm intervals. The results showed that The PLS calculation for Dextromethorphan HBr has 110.07% and Glyceryl Guaiacolate was 110.01% with FTIR methods and Spectrophotometric methods were 104.73 % for Dextromethorphan HBr and 101.78 for Glyceryl Guaiacolate. The results showed that the simultaneous determination of the mixture of dextromethorphan HBr and Glyceryl Guaiacolate utilizing FTIR, and spectrophotometric methods with PLS calculation can be used and meets the validation requirements.

The Use of Chemometrics for Classification of Sidaguri (&lt;i&gt;Sida rhombifolia&lt;/i&gt;) Based on FTIR Spectra and Antiradical Activities

Sidaguri (Sida rhombifolia) is one of the herbal components used in traditional medicine. The application of chemometrics in the standardization of herbal medicine is common. The objective of this study was to classify Sidaguri from different regions based on FTIR spectra with chemometrics of principal component analysis (PCA) and to correlate the antioxidant activities with FTIR spectra using the multivariate calibration of partial least square regression (PLSR). The extraction of Sidaguri powder was performed using ultrasound-assisted extraction (UAE) at optimum conditions. The obtained extracts were subjected to antiradical scavenging activities using DPPH (2,2’-diphenyl-1-picrylhydrazyl) and ABTS (2,2′-azinobis-3-ethylbenzothiazoline-6-sulfonic acid) radicals. The PCA result shows that Sidaguri from different regions could be separated using 14 wavenumbers of FTIR spectra based on the PCA's loading plot. PLSR regression using the second derivative FTIR spectra at wavenumbers of 3662–659 cm–1 could predict radical scavenging activities (RSA) of Sidaguri with R2 values of 0.9636 and 0.9024 for calibration and validation models, with RMSEC and RMSEP values of 1.45% and 2.65%, respectively. It can be concluded that FTIR spectra treated by PCA were reliable for classifying Sidaguri from different regions. At the same time, PLSR was accurate and precise enough to predict the RSA of Sidaguri.

Determination of extractive content in Cupressus sempervirens wood through a NIRS-PLSR model and its correlation with durability

The objective of this study was the development of a NIRS (Near InfraRed Spectroscopy) and PLSR (Partial Least Square Regression) model for quantitatively determining extractives, in four clones of the Italian cypress Cupressus sempervirens. Extractives were correlated to the fungal resistance of wood samples evaluated with a non-standard procedure based on the mass loss of mini-blocks. PLSR regression models based on NIRS gave good predictions of extractives content with RPD (Residual Prediction Deviation) equal or above 2.7, suggesting that NIRS-PLSR procedure could be used for the characterization of cypress wood. Total extractives were significantly different among heartwood and sapwood but not among clones. Both inner and outer heartwood of the four C. sempervirens clones were resistant to three brown rot fungi. Minimal differences were found among clones in the mass loss of the heartwood samples, while significant inter-clonal differences emerged in the mass loss of sapwood samples. Intra-clonal variation was detected for sapwood samples and only in one of the four tested clones. The correlation between durability and extractives content was significant only if sapwood was added to heartwood in the dataset. Nevertheless, the high durability observed in the examined clones encourages selection of these cypress genotypes for wood production.

Synthesis, biological activity, and four-dimensional quantitative structure–activity analysis of 2-arylidene indan-1,3-dione derivatives tested against Daphnia magna

ABSTRACT A series of 18 2-arylidene indan-1,3-dione derivatives was synthesized and tested against Daphnia magna to assess the environmental toxicity of these compounds. Aiming to investigate the toxicity mechanism for this series of compounds, a four-dimensional quantitative structure–activity analysis (4D-QSAR) was performed through the partial least square regression (PLS). The best PLS model was built with two factors and the selected field descriptors, of Coulomb (C) and Lennard-Jones (L) nature, describing 77.43% of variance and presenting the following statistics: r 2 = 0.89; SEC = 0.30; Q 2 = 0.81; SEV = 0.36. According to the literature, the bioactivity of α,β-unsaturated ketones, a functionality present in the series of compounds under investigation, is related to the conjugated double bond with the carbonyl group. The presence of a positive Coulomb descriptor nearby the carbonyl moieties, obtained as a result of the regression model, indicates that these polar groups are also related to the toxicity on D. magna. From the PLS regression model, the toxicity EC50-48 h values increases with the positive Coulomb descriptor and diminishes with the negative Lennard-Jones descriptors. It could be concluded that the presence of small polar groups in the aromatic ring of the arylidene moiety tends to increase the toxicity, while bulkier apolar substituents lead to a decrease of the toxicity.

A green and high throughput method for salt determination in crude oil using digital image-based colorimetry in a portable device

A digital image-based (DIB) method using colorimetry via a portable device is proposed for salt determination in crude oil. Chloride was extracted to an aqueous phase, followed by colorimetric quantification using digital images. A USB camera was attached to a homemade 3D-printed chamber with controlled light intensity. The images were converted into red, green, and blue (RGB) histograms, and a partial least squares regression (PLS) model was constructed using the free PhotoMetrix UVC app. The experimental parameters were optimized, as the ratio between silver nitrate and the potassium chromate, the lighting, the volume of solution in the measurement chamber, the focal distance, and the pH effect. A miniaturized extraction method was developed for sample preparation, and the main parameters were evaluated: sample mass, time, temperature, solvent, or the use of a demulsifier and its amount. The PLS regression model was evaluated in terms of the coefficient of determination (R2), the root mean squared error of calibration (RMSEC), the root mean squared error of cross validation (RMSECV) and the root mean squared error of prediction (RMSEP). The detection limit was 69 mg L−1 (0.034 mg g−1 using 10 g of sample). No differences were observed (p < 0.05) when comparing the results from the DIB method with the reference values. The extraction method’s simplicity combined with the digital image-based colorimetric analysis allowed an easy-to-use and environmentally friendly method. Other important advantages include low cost, the possibility to be portable, and high throughput in comparison to standard methods.

Raman Spectroscopy as a fast tool for whey quantification in raw milk

Adulteration of raw milk with whey is difficult to evaluate because both have great physical and chemical similarities. Herein, we proposed a rapid and cost-effective method to detect and quantify the presence of whey in milk using Raman spectroscopy, while dispensing any step of sample preparation. Principal Component Analysis (PCA) was performed for outlier detection and Partial Least Squares regression (PLS) for prediction of whey content. Since single vibrational modes have been not identified in order to discriminate the different levels of whey addition, we proposed a calibration model from the full Raman spectra using PLS regression. We explored the idea that small differences in spectroscopic data could be useful in this evaluation by using multivariate chemometric methods. As result, the correlation coefficient between the Raman method and the experimental values was R² = 0.988 for the validation samples. A paired t-test revealed a p-value greater than 0.5, pointing out that no significant statistical differences between the proposed Raman method and the reference values. The approach proposed in this study shows to be quite useful for the evaluation of milk adulteration by whey addition, thus contributing to quality control of milk in dairy industry.

Could antioxidant capacity and flavonoid content of ethanolic extracts of geopropolis from Brazilian native bees be estimated from digital photos and NIR Spectra?

In the current study, Multi Imaging Analysis (MIA) of digital photographs and NIR spectra of ethanolic extracts of geopropolis (EEGs) from Brazilian native bees were used for the first time to build partial least square regression (PLS) models to predict the antioxidant capacity (AC) and Total Flavonoid Content (TFC) of EEGs. Firstly, spectrophotometric reference assays were carried out: TFC was determined by complexation with AlCl3 and AC by the antiradical scavenging of the stable radical 2,2-diphenyl-1-picrylhydrazine (DPPH), Ferric Reducing Antioxidant Power (FRAP) and Cupric Ion Reducing Antioxidant Capacity (CUPRAC). Subsequently, NIR spectra and digital photos of EEGs were taken and the raw data was calibrated using PLS regression. All Color-based and NIR-based PLS models were appropriate to estimate AC and TFC in EEGs but the best performance and precision were obtained using Color-based PLS models (R2 for calibration ≥0.74 together with low RMSEC and RMSECV and lesser latent variables). The predictive capacity of Color-based and NIR-based PLS models was similar, although MIA-TFC and MIA-DPPH models indicated more reliability and accurate predictive capacity (R2 > 0.80 together with low RMSEP). Consequently, simple digital photos of geopropolis extracts or NIR spectra combined with PLS models can be potentially used to predict the total flavonoid content and antioxidant capacity of geopropolis with acceptable accuracy. This approach might be employed in the control quality of geopropolis extracts with the advantages of being a fast, low cost and green methodology.

High-performance liquid chromatography with fluorescence detection fingerprinting combined with chemometrics for nut classification and the detection and quantitation of almond-based product adulterations

Economically motivated food fraud has increased in recent years, with adulterations and substitutions of high-quality products being common practice. Moreover, this issue can affect food safety and pose a risk to human health by causing allergies through nut product adulterations. Therefore, in this study, high-performance liquid chromatography with fluorescence detection (HPLC-FLD) fingerprints were used for classification of ten types of nuts, using partial least squares regression-discriminant analysis (PLS-DA), as well as for the detection and quantitation of almond-based product (almond flour and almond custard cream) adulterations with hazelnut and peanut, using partial least squares regression (PLS). A satisfactory global nut classification was achieved with PLS-DA. Paired PLS-DA models of almonds in front of their adulterants were also evaluated, producing a classification rate of 100%. Moreover, PLS regression produced low prediction errors (below 6.1%) for the studied adulterant levels, with no significant matrix effect observed.

Comprehensive characterisation of ylang-ylang essential oils according to distillation time, origin, and chemical composition using a multivariate approach applied to average mass spectra and segmented average mass spectral data

Analyses of the complex essential oil samples using gas chromatography hyphenated with mass spectrometry (GC–MS) generate large three-way data arrays. Processing such large data sets and extracting meaningful information in the metabolic studies of natural products requires application of multivariate statistical techniques (MSTs). From the GC–MS raw data several different input data sets for the MSTs can be created, including total chromatogram average mass spectra (TCAMS), segmented average mass spectra (SAMS) and chemical composition. Herein, we compared the performance of MSTs on average mass spectrum based data sets, TCAMS and SAMS, against chemical composition and attenuated total reflectance - Fourier transformation infrared (ATR-FTIR) spectroscopy in the evaluation of quality of ylang-ylang essential oils, based on their grade, geographical origin and chemical composition, using principal component analysis (PCA), partial least squares regression (PLS) and discriminatory analysis (PLS-DA). PCA based on TCAMS, SAMS and chemical composition showed clear trends amongst the samples based on increase in grade (distillation time). PLS-DA applied to TCAMS, SAMS and ATR-FTIR discriminated between all geographical origins. Predicted relative abundances of the 18 most important compounds, using PLS regression models on TCAMS, SAMS and ATR-FTIR, were successfully applied to ylang-ylang essential oil quality assessment based on comparison with the ISO 3063:2004 standard, where the SAMS data set showed superior performance, compared to other data sets.

An evaluation model for the deep dewatering potential of municipal sludge based on PLS regression

For constructing a new convenient and reliable system to evaluate the dewatering performance of sludge, the relationship between the water content of dewatered municipal sludge and its physicochemical properties was comprehensively investigated. Then, an evaluation and prediction model was established based on partial least squares regression (PLS), which could accurately predict the solid content of municipal sludge treated by high-pressure deep dewatering. Furthermore, the main factors affecting the sludge dewaterability were explored by the means of principal component analysis (PCA), and it was concluded that first principal component (PC1), containing most composition of extracellular polymeric substances (EPS), accounted for 53.1% of the X variation and 62.9% of information Y variation. Both protein in soluble EPS (PNS) and polysaccharide in a tightly bound EPS (PST), with the variable importance in the projection (VIP) values more than 1, were regarded as the most decisive factors to the deep dewatering performance of municipal sludge. Additionally, the validation results demonstrate the accuracy and practicability of the obtained model, specifically fit for a conditioning system for cationic polyacrylamide (CPAM) and lysozyme either separately or jointly.

Estimating Chlorophyll Content of Leafy Green Vegetables from Adaxial and Abaxial Reflectance.

As a primary pigment of leafy green vegetables, chlorophyll plays a major role in indicating vegetable growth status. The application of hyperspectral remote sensing reflectance offers a quick and nondestructive method to estimate the chlorophyll content of vegetables. Reflectance of adaxial and abaxial leaf surfaces from three common leafy green vegetables: Pakchoi var. Shanghai Qing (Brassica chinensis L. var. Shanghai Qing), Chinese white cabbage (Brassica campestris L. ssp. Chinensis Makino var. communis Tsen et Lee), and Romaine lettuce (Lactuca sativa var longifoliaf. Lam) were measured to estimate the leaf chlorophyll content. Modeling based on spectral indices and the partial least squares regression (PLS) was tested using the reflectance data from the two surfaces (adaxial and abaxial) of leaves in the datasets of each individual vegetable and the three vegetables combined. The PLS regression model showed the highest accuracy in estimating leaf chlorophyll content of pakchoi var. Shanghai Qing (R2 = 0.809, RMSE = 62.44 mg m−2), Chinese white cabbage (R2 = 0.891, RMSE = 45.18 mg m−2) and Romaine lettuce (R2 = 0.834, RMSE = 38.58 mg m−2) individually as well as of the three vegetables combined (R2 = 0.811, RMSE = 55.59 mg m−2). The good predictability of the PLS regression model is considered to be due to the contribution of more spectral bands applied in it than that in the spectral indices. In addition, both the uninformative variable elimination PLS (UVE-PLS) technique and the best performed spectral index: MDATT, showed that the red-edge region (680–750 nm) was effective in estimating the chlorophyll content of vegetables with reflectance from two leaf surfaces. The combination of the PLS regression model and the red-edge region are insensitive to the difference between the adaxial and abaxial leaf structure and can be used for estimating the chlorophyll content of leafy green vegetables accurately.

Evaluation of Leaf N Concentration in Winter Wheat Based on Discrete Wavelet Transform Analysis

Leaf nitrogen concentration (LNC) is an important indicator for accurate diagnosis and quantitative evaluation of plant growth status. The objective was to apply a discrete wavelet transform (DWT) analysis in winter wheat for the estimation of LNC based on visible and near-infrared (400–1350 nm) canopy reflectance spectra. In this paper, in situ LNC data and ground-based hyperspectral canopy reflectance was measured over three years at different sites during the tillering, jointing, booting and filling stages of winter wheat. The DWT analysis was conducted on canopy original spectrum, log-transformed spectrum, first derivative spectrum and continuum removal spectrum, respectively, to obtain approximation coefficients, detail coefficients and energy values to characterize canopy spectra. The quantitative relationships between LNC and characteristic parameters were investigated and compared with models established by sensitive band reflectance and typical spectral indices. The results showed combining log-transformed spectrum and a sym8 wavelet function with partial least squares regression (PLS) based on the approximation coefficients at decomposition level 4 most accurately predicted LNC. This approach could explain 11% more variability in LNC than the best spectral index mSR705 alone, and was more stable in estimating LNC than models based on random forest regression (RF). The results indicated that narrowband reflectance spectroscopy (450–1350 nm) combined with DWT analysis and PLS regression was a promising method for rapid and nondestructive estimation of LNC for winter wheat across a range in growth stages.

Identification of phenolic and alcoholic compounds in wine spirits and their classification by use of multivariate analysis

During the ageing period wine spirits are changing their color, chemical composition and sensory characteristics. These changes should be simply monitored. The aim of this study was to develop partial least squares regression (PLS) models for higher alcohols and phenols in wine spirits as well as to show the feasibility of the NIR spectroscopy combined with chemometric tools to distinguish wine spirits and brandies with different ageing degree. To get the reference values, the usual methods for the analysis of spirits drinks were used. Ethanol, esters, acids, methanol and higher alcohols were studied. Wine spirits and brandies phenol composition was determined by liquid chromatography. Principal component analysis (PCA) was used to classify the wine spirits and brandies according to their phenolic and higher alcohols composition. Moreover, the Partial least squares regression (PLS regression) was used to calibrate and predict expected contents of higher alcohols and phenols in the wine spirits. Success of the classification of samples by ageing based on individual alcohols was 93.8 %, while success of the classification based on individual phenols raised to 100 %. This efficiency of the prediction was evaluated by use of linear discriminator analysis (LDA).

Determinants behaviour analysis of demand for passengers to the Portuguese air travel leisure market: Application with Partial Least Squares Regression (PLS)

This paper aims to determine and analyse the behaviour of the demand for passengers' leisure air travel in the Portuguese market (between Lisbon and Oporto and the Azores and Madeira), between 2010 and 2014, using an application of partial least squares regression (PLSR). The PLS regression analysis used in the essay was developed by Wold (1985) and in accordance with Tenenhaus et al.'s (2005) study. In the first section we present a contextualisation of the PLSR. In a second section we present the data and the model. In a third section we present the discussion of results. Finally, we present the final considerations. The results of our empirical analysis suggest that the demand for air transport of passengers between Portugal and the Azores and Madeira is more affected by the destination attraction factors than by aviation industry factors. The results were promising and provide important information to improve business decision making.

A comparison of non-linear regression methods for improved on-line near infrared spectroscopic analysis of a sugarcane quality measure

On-line near infrared (NIR) spectroscopic analysis systems play an important role in assessing the quality of sugarcane in Australia. As quality measures are used to calculate the payment made to growers, it is imperative that NIR models are both accurate and robust. Machine learning and non-linear modelling approaches have been explored as methods for developing improved NIR models in a variety of industrial settings, yet there has been little research into their application to cane quality measures. The objective of this paper was to compare chemometric models of commercial cane sugar (CCS) based on four calibration techniques. CCS was estimated using partial least squares regression (PLS), support vector regression (SVR), artificial neural networks (ANNs) and gradient boosted trees (GBTs). Model performance was assessed on an independent validation data set using root mean square error of prediction (RMSEP) and r2 values. SVR (RMSEP = 0.37%; r2 = 0.92) and ANN (RMSEP = 0.36%; r2 = 0.93) performed similarly to PLS (RMSEP = 0.37%; r2 = 0.92) on the validation data set, while GBT exhibited a much lower skill (RMSEP = 0.51%; r2 = 0.85). Analysis of important wavelengths in each model showed that PLS regression, SVR and ANN techniques emphasized the importance of similar spectral regions. Future research should consider testing model robustness over seasons and/or regions. Comparisons of chemometric models should consider reporting variable importance as a way of understanding how models use spectral information.