Determination of the Polar Compounds in Vegetable Oil by Ultra-Performance Liquid Chromatography–Quadrupole-Time-of-Flight-Mass Spectrometry with Chemometrics

High-throughput ultra-performance liquid chromatography–quadrupole time-of-flight mass spectrometry was combined with chemometric tools for the rapid determination of polar components in camellia oil, rapeseed oil, and waste cooking oil. The results were analyzed by two unsupervised methods: principal component analysis (one-way ANOVA, p<.05) and volcano plot analysis (p<.05, fold change ≥2) and supervised method: partial least squares discriminant analysis. The results showed that the oils were correctly classified based on their polar components. The first three principal components reflected most detail with a cumulative contribution rate of 84.67% using principal component analysis. The prediction accuracy was close to 100% using partial least squares discriminant analysis. Nineteen components were screened by principal component analysis; twelve were preliminary identified as palmitamide, phytosphingosine, eicosasphinganine, 1-monopalmitin, glyceryl monooleate, glyceryl monostearate, 1α-hydroxyvitamin D2, 1-linoleoyl glycerol, oleamide, sphinganine, stearamide, and linoleic acid. The proposed method may be applied to effectively and accurately authenticate edible oils.     

基于多元统计学的青稞HPLC指纹图谱研究

为了筛选品质良好的青稞,选择不同产地的10批青稞样品,建立了青稞的高效液相指纹图谱,并进行了聚类分析和主成分分析,测定了儿茶素、表儿茶素和芦丁3种指标成分的含量.首先采用高效液相色谱法测得了10批青稞的色谱图,建立其共有模式,确定了11个共有峰,蓝青稞指标成分含量均高于白青稞.对共有峰数据进行聚类分析和主成分分析,10...     

Chemometrical interpretation of analytical data for vertical distribution of some chemical components in sea water

Summary Analytical data for vertical distribution (depth profile) of components in sea water are interpreted by the use of multivariate statistical methods (principal components analysis, correspondence factor analysis and cluster analysis). The various chemical components (NH₃, P, Zn, Mn, Cu, Fe, O₂, pH, salinity and alkalinity) and some physical characteristics (E_h, T, suspended matter) or the depths of sampling (from 0 to 200 m) are classified as objects of similarity. Two principal components are identified as possible really existing natural factors determining the vertical distribution of the components in the sea water phase.     

主成分分光光度法中主成分的选择

主成分分析是全光谱分析度分析中常用的校正方法。本文提出第一主成分并不是与因最线性相关的主成分。为此，我们利用扫描算法众多主成分中选择与因变量（浓度）最相关的主成分，从而使计算结果更准确可信。本文还对单因变量和多因变量两种情况下主成分选择的统计量进行了讨论。     

Multivariate monitoring of fermentation processes with non-linear modelling methods

Multiway principal components analysis (MPCA) and parallel factor analysis (PARAFAC) are widely used in exploratory data analysis and multivariate statistical process control (MSPC). These models are linear in nature, thus, limited when non-linear relations are present in the data. Principal component analysis (PCA) can be extended to non-linear principal components analysis using autoassociative neural networks. In this paper, the network’s bottleneck layer outputs (non-linear components) were made orthogonal. A method to estimate confidence limits based on a kernel probability density function was proposed since these limits do not assume that the non-linear scores are normally distributed. A measure for the non-linear scores (D_NL) was presented here to monitor on-line the process replacing the well known Hotelling’s T² statistic. One hundred and two industrial fermentation runs were used to evaluate the performance of a non-linear technique for multivariate process statistical monitoring. Three process runs with faults were used to compare the error detection performance using a statistic for the non-linear scores and the residuals statistic (SPE).     

Multivariate statistical assessment of polluted soils

This study deals with the application of several multivariate statistical methods (cluster analysis, principal components analysis, multiple regression on absolute principal components scores) for assessment of soil pollution by heavy metals. The sampling was performed in a heavily polluted region and the chemometric analysis revealed four latent factors, which describe 84.5 % of the total variance of the system, responsible for the data structure. These factors, whose identity was proved also by cluster analysis, were conditionally named “ore specific”, “metal industrial”, “cement industrial”, and “steel production” factors. Further, the contribution of each identified factor to the total pollution of the soil by each metal pollutant in consideration was determined.     

基于主成分分析和小波神经网络的近红外多组分建模研究

将小麦叶片原始光谱经过预处理后,采用主成分分析(PCA)对数据进行降维,取前3个主成分输入小波神经网络,建立了基于主成分分析和小波神经网络的近红外多组分预测模型(WNN);进一步研究了小波基函数个数的选取(WNN隐层节点数)对小波神经网络模型性能的影响,并将WNN模型与偏最小二乘法(PLS)和传统的反向传播神经网络(BPNN)模型进行了比较.结果表明,所建立的WNN模型能用于同时预测小麦叶片全氮和可溶性总糖两种组分含量,其预测均方根误差(RMSEP)分别为0.101%和0.089%,预测相关系数(R)分别为0.980和0.967.另外,在收敛速度和预测精度上,WNN模型明显优于BPNN和PLS模型,从而为将小波神经网络用于近红外光谱的多组分定量分析奠定了基础.     

Analysis of Psoralea corylifolia L. fruits in different regions

Application of multivariate data analysis has become a popular method in the last decades, mainly because it can provide information not otherwise accessible. The information includes classification, searching similarities, finding relationships, finding physical significance to principal components, etc. Twenty-two Chinese medicinal herbs containing twelve constituents were collected and determined by HPLC. The results were studied by hierarchical cluster analysis (HCA) and principal components analysis (PCA). It was shown that the samples could be clustered reasonably into three groups, hence corresponding with the typical habitats of Psoralea corylifolia L.     

Classification of byzantine glass samples by principal component analysis

Summary Nine samples of byzantine glass classified previously by cluster analysis are classified by principal component analysis (PCA). A visual inspection of plots in coordinates of the first two principal components gives essentially the same results as cluster analysis. In addition, PCA indicates relationships among the classification variables.

---

Klassifizierung byzantinischer Glasproben durch Analyse der Hauptbestandteile

     

基于地统计学与支持向量回归的QSAR建模

基于主成分分析(PCA)、地统计学(GS)和支持向量回归(SVR), 提出了一种新的定量构效关系(QSAR)个体化预测方法——Weight-PCA-GS-SVR. 其基本思路是: 先以PCA降维并消除自变量间的信息冗余, 继以SVR经非线性主成分筛选去除与因变量无关的主成分, 再以保留主成分计算样本间的加权距离, 然后以高维GS确定公用变程; 每一个待测样本都以自身为中心从训练集中找出加权距离小于公用变程的私有k个近邻, 以SVR训练建模完成个体化预测. Weight-PCA-GS-SVR从行、列两个方向对模型进行了优化, 为自变量提供了一种新的加权方法, 为解决最优k近邻选择难题提供了新的思路, 并具有SVR原来的优点. 经3个化合物活性实例数据集验证, 新方法在所有参比模型中预测精度最高, 且明显优于文献报道结果, Weight-PCA-GS-SVR在QSAR等回归预测领域有较广泛的应用前景.     

Rapidly differentiating grape seeds from different sources based on characteristic fingerprints using direct analysis in real time coupled with time‐of‐flight mass spectrometry combined with chemometrics

The seeds of grapevine (Vitis vinifera) are a byproduct of wine production. To examine the potential value of grape seeds, grape seeds from seven sources were subjected to fingerprinting using direct analysis in real time coupled with time‐of‐flight mass spectrometry combined with chemometrics. Firstly, we listed all reported components (56 components) from grape seeds and calculated the precise m/z values of the deprotonated ions [M–H]^–. Secondly, the experimental conditions were systematically optimized based on the peak areas of total ion chromatograms of the samples. Thirdly, the seven grape seed samples were examined using the optimized method. Information about 20 grape seed components was utilized to represent characteristic fingerprints. Finally, hierarchical clustering analysis and principal component analysis were performed to analyze the data. Grape seeds from seven different sources were classified into two clusters; hierarchical clustering analysis and principal component analysis yielded similar results. The results of this study lay the foundation for appropriate utilization and exploitation of grape seed samples. Due to the absence of complicated sample preparation methods and chromatographic separation, the method developed in this study represents one of the simplest and least time‐consuming methods for grape seed fingerprinting.     

有机磷农药构效关系的主成分分析-人工神经网络研究

采用主成分分析法对样本数据集进行预处理，将得到的新的样本数据集输入人工神经网络，对有机磷农药的毒性参数进行预报。研究结果表明，主成分分析－人工神经网络的预报精度优于单纯的人工神经网络。     

A cautionary tale of principal component analysis: an example from inductively-coupled plasma/atomic emission spectrometry

Principal component analysis (PCA) is a widely used chemometric technique, but there can be serious limitations on the validity of the conclusions. In the example given, PCA was applied to identify the causes of noise and drift in inductively-coupled plasma/atomic emission spectrometry (ICP/AES). The effects of ten possible instrumental variables on lines of 24 elements were measured independently to establish distinctive “fingerprints”. PCA showed that >90% of the variance in routine analysis was correlated between elements, rather than being random. However, the multi-element ‘fingerprints’ identified as the principal components did not correspond to any of those established as suspected causes. A model based on variation of two of the suspected causes of error was used to simulate analysis by ICP/AES. When subjected to PCA, the simulated data also gave erroneous fingerprints in the principal components that closely matched those found in the real data. The reason for the failure of PCA to identify the causes of the variation was found to be that the two apparently independent causes had correlated multi-element effects. Caution is therefore required in the identification of principal components in terms of recognisable features.     

Supervised principal components: a new method for multivariate spectral analysis

The supervised principal components (SPC) method was proposed by Bair and Tibshirani for statistics regression problems where the number of variables greatly exceeds the number of samples. This case is extremely common in multivariate spectral analysis. The objective of this research is to apply SPC to near‐infrared and Raman spectral calibration. SPC is similar to traditional principal components analysis except that it selects the most significant part of wavelength from the high‐dimensional spectral data, which can reduce the risk of overfitting and the effect of collinearity in modeling according to a semi‐supervised strategy. In this study, four conventional regression methods, including principal component regression, partial least squares regression, ridge regression, and support vector regression, were compared with SPC. Three evaluation criteria, coefficient of determination (R²), external correlation coefficient (Q²), and root mean square error of prediction, were calculated to evaluate the performance of each algorithm on both near‐infrared and Raman datasets. The comparison results illustrated that the SPC model had a desirable ability of regression and prediction. We believe that this method might be an alternative method for multivariate spectral analysis. Copyright © 2013 John Wiley & Sons, Ltd.     

Cross‐validation in PCA models with the element‐wise k‐fold (ekf) algorithm: theoretical aspects

Cross‐validation has become one of the principal methods to adjust the meta‐parameters in predictive models. Extensions of the cross‐validation idea have been proposed to select the number of components in principal components analysis (PCA). The element‐wise k‐fold (ekf) cross‐validation is among the most used algorithms for principal components analysis cross‐validation. This is the method programmed in the PLS_Toolbox, and it has been stated to outperform other methods under most circumstances in a numerical experiment. The ekf algorithm is based on missing data imputation, and it can be programmed using any method for this purpose. In this paper, the ekf algorithm with the simplest missing data imputation method, trimmed score imputation, is analyzed. A theoretical study is driven to identify in which situations the application of ekf is adequate and, more importantly, in which situations it is not. The results presented show that the ekf method may be unable to assess the extent to which a model represents a test set and may lead to discard principal components with important information. On a second paper of this series, other imputation methods are studied within the ekf algorithm. Copyright © 2012 John Wiley & Sons, Ltd.     

Principal component analysis of dipeptides

Principal component analysis applied to a set of dipeptides illustrates how changes in families of parameters act in concert to produce overall molecular structural changes. Principal component analysis is an eigenvalue–eigenvector analysis whereby the parametric sensitivity coefficient matrix is manipulated to produce weighted principal components, which reveal the variant and invariant directions in the parameter space. This analysis summarizes the sensitivity results by revealing interdependence among the parameter values with regard to their role in controlling the molecular structure. An analysis of the principal components reveals hidden relationships among the parameters. Thus, those parameters, which were thought to be of controlling significance with respect to the molecular structure, may, in fact, not be (or vice versa) due to cooperative parametric interactions; as a result, the parameters of significance in a sequence of dipeptides are identified. In general, for the dipeptides studied, there is mutual exclusion of dominant parameters between the sets of invariant and variant eigenvectors. © 1994 by John Wiley & Sons, Inc.     

Principal components analysis for the estimation of interdependences among trace metals in cow milk

Seven trate metals (Cr, Mn, Fe, Ni, Cu, Cd, Pb) were determined in the dissolved ash of pasteurized milk by electrothermal atomic absorption spectrometry. The distribution of the concentration (c) of the metals in milk of χ = log₁₀c for eaeh metal was investigated by means of the Lin-Mudholkar test for normality. Logarithmically transformed variables were first considered for further processing, as their distribution passed the normality test at a 0.05 significance level for a sample size of 48. The correlation matrix around the mean was used as a starting matrix for principal components analysis; the principal components were obtained from the FACTOR program of the SPSS package. Dimensions were reduced to four principal components, accounting for 78% of the total variance. Various orthogonal rotations indicated associations with Cd- Pb, Cr- Ni, and Mn- Cu. The correlation matrix was also estimated from the c matrix after row- normalization for each sample. The first eigenvalue estimated from this matrix accounted for 50% of the total variance, bur three eigenvalues were needed to reach 80% of explained variance. Cadmium and lead formed a cluster of variables, indicating a common origin. Features concerned with natural metal contents and contamination during transport and processing are diseussed.     

High‐performance liquid chromatography based chemical fingerprint analysis and chemometric approaches for the identification and distinction of three endangered Panax plants in Southeast Asia

Among Panax genus, only three endangered species Panax notoginseng, P. vietnamensis, and P. stipuleanatus that have a similar morphology are mainly distributed in Southeast Asia. These three plants are usually misidentified or adulterated. To identify them well, their chemical chromatographic fingerprints were established by an effective high‐performance liquid chromatography method. By comparing the chromatograms, the three Panax species could be distinguished easily using the 22 characteristic peaks. Besides, the data of the chromatographic fingerprints aided by chemometric approaches were applied for the identification and investigation the relationship of different samples and species. Using similarity analysis, the chemical components revealed higher similarity between P. vietnamensis and P. stipuleanatus. The results of hierarchical clustering analysis indicated that samples belonging to the same species could be clustered together. The result of principal component analysis was similar with hierarchical clustering analysis and the three principal components accounted for >80.5% of total variability.