Journal of Chemometrics最新文献

Deep learning algorithms represented by convolutional neural networks bring new opportunities for spectral analysis technology. Convolutional neural networks are more straightforward than traditional chemometric algorithms for detecting the quality of agricultural products, reducing the procedures of spectral preprocessing and band selection, and with higher prediction accuracy. However, there are few research papers on the relevance of the explanation of the convolutional neural networks model mechanism, and the reader cannot fully understand convolutional neural networks feature learning. In this study, convolutional neural networks combined with the subregional voting method were used to predict the moisture content of semen ziziphi spinosae. Firstly, 10 regions of interest were divided using the subregional voting method, and the results of network models were compared. It was found that the average spectrum of the fifth region of interest had the best prediction of moisture content because it was closest to the central region of semen ziziphi spinosae. Based on this, a convolutional neural network containing three convolutional layers, three pooling layers, and one fully connected layer is proposed. Partial least squares regression, backpropagation neural network, and convolutional neural networks were established to predict the moisture content of semen ziziphi spinosae. The correlation coefficient of the prediction set of the partial least squares regression is 0.98 after the multiplicative scatter correction preprocessed the spectra, and correlation coefficient of the prediction set of the backpropagation neural network is 0.83 after the standard normal variate preprocessed the spectra. The correlation coefficient of the prediction set of the convolutional neural networks established by using the raw spectra reached 0.99. The spectral preprocessing method can improve the prediction set correlation coefficient of partial least squares regression and backpropagation neural network. Still, it will reduce the prediction ability of convolutional neural networks. This study also analyzed the effect of different learning rates on the performance of convolutional neural networks, and it was found that the training loss and training accuracy performed most consistently when the learning rate was 0.01. Secondly, this study also visualized the output feature maps of the three convolutional layers of convolutional neural networks and verified the effectiveness of convolutional neural networks feature band extraction. This study provides new ideas for deep learning in the online detection of seed moisture content.

Multi-way data analysis is popular in chemometrics for the decomposition of, for example, spectroscopic or chromatographic higher-order tensor datasets. Parallel factor analysis (PARAFAC) and its extension, PARAFAC2, are extensively employed methods in chemometrics. Applications of PARAFAC2 for untargeted data analysis of hyphenated gas chromatography coupled with mass spectrometric detection (GC-MS) have proven to be very successful. This is attributable to the ability of PARAFAC2 to account for retention time shifts and shape changes in chromatographic elution profiles. Despite its usefulness, the most common implementations of PARAFAC2 are considered quite slow. Furthermore, it is difficult to apply constraints (e.g., non-negativity) to the shifted mode in PARAFAC2 models. Both aspects are addressed by a new shift-invariant tri-linearity (SIT) algorithm proposed in this paper. It is shown on simulated and real GC-MS data that the SIT algorithm is 20–60 times faster than the latest PARAFAC2-alternating least squares (ALS) implementation and the PARAFAC2-flexible coupling algorithm. Further, the SIT method allows the implementation of constraints in all modes. Trials on real-world data indicate that the SIT algorithm compares well with alternatives. The new SIT method achieves better factor resolution than the benchmark in some cases and tends to need fewer latent variables to extract the same chemical information. Although SIT is not capable of modeling shape changes in elution profiles, trials on real-world data indicate the great robustness of the method even in those cases.

Chemists have been pursuing general mathematical laws to explain and predict molecular properties for a long time. However, most of the traditional quantitative structure-activity relationship (QSAR) models have limited application domains; for example, they tend to have poor generalization performance when applied to molecules with parent structures different from those of the trained molecules. This paper attempts to develop a new QSAR method that is theoretically possible to predict various properties of molecules with diverse structures. The proposed deep electron cloud-activity relationships (DECAR) and deep field-activity relationships (DFAR) methods consist of three essentials: (1) a large number of molecule entities with activity data as training objects and responses; (2) three-dimensional electron cloud density (ECD) or related field data by the accurate density functional theory methods as input descriptors; and (3) a deep learning model that is sufficiently flexible and powerful to learn the large data described above. DECAR and DFAR are used to distinguish 977 sweet and 1965 non-sweet molecules (with 6-fold data augmentation), and the classification performance is demonstrated to be significantly better than the traditional least squares support vector machine (LS-SVM) models using traditional descriptors. DECAR and DFAR would provide a possible way to establish a widely applicable, cumulative, and shareable artificial intelligence-driven QSAR system. They are likely to promote the development of an interactive platform to collect and share the accurate ECD and field data of millions of molecules with annotated activities. With enough input data, we envision the appearance of several deep networks trained for various molecular activities. Finally, we could anticipate a single DECAR or DFAR network to learn and infer various properties of interest for chemical molecules, which will become an open and shared learning and inference tool for chemists.

The determination of nitrite concentration is crucial due to its toxicity. A novel model has been developed to accurately determine nitrite concentration within the non-linear range, utilizing the Zambelli method. Previously, techniques for measure nitrite concentration were primarily restricted to the linear range. This new method employs UV-Visible absorption spectra and correlated component regression (CCR) to determine nitrite concentration within the range of 0.27–11.34 ppm. A wavelength selection strategy in conjunction with partial least squares (PLS) was implemented prior to applying CCR. The spectral data underwent pre-processing using standard normal variant (SNV) and Savitzky Golay (SG) techniques, and a backward selection (BS) strategy with PLS was applied to select wavelengths. The 15 most sensitive wavelengths, determined through the RMSE_CV criterion, were utilized to create a PLS model within the range 377–497 nm, resulting in a model with R²_C = 0.9999 and R²_CV = 0.9999, RMSE_C = 0.006, RMSE_CV = 0.027. A CCR model was then established using the 15selected wavelengths and nitrite concentration. The results yielded strong correlation between predicted and measured nitrite values with R²_C = 0.9996, RMSE_C = 4.7491 E-15, RMSE_CV = 0.0004, and MAPE = 0.68%. The method has been validated through an accuracy profile, which demonstrates that 80% of future results will fall within the 10% acceptability limit within the validation range of 1.30–8.83 mg/L.

The Journal of Chemometrics is pleased to announce a special issue focused on multivariate analysis of data from designed experiments. ANOVA (Analysis of Variance) is the standard method for analyzing data from experimental designs. The classical ANOVA methods are however univariate and do not handle multiple collinear response variables. Designed experiments with multivariate outputs are prevalent across various scientific disciplines, necessitating methods that appropriately consider both the experimental design and the multivariate nature of the data.

Several multivariate ANOVA techniques have been presented already. The most prevalent approaches involve combining ANOVA with PCA (principal component analysis) or other exploratory component-based techniques in different ways. Some commonly used methods in this context include ASCA, ANOVA-PCA, AComDim, and fifty-fifty MANOVA. These methods integrate ANOVA and PCA in different ways to extract meaningful information from multivariate data. Additionally, there are alternative methods that replace PCA with partial least squares (PLS) regression, which allows for the utilization of PLS-specific validation and variable importance routines. One major advantage of all these methods is that they not only offer interpretation and variable importance metrics from latent variable-based methods but also provide estimates of multivariate effect sizes accompanied by corresponding significance testing.

Despite the progress made in recent years, the field of multivariate analysis of data from designed experiments is still young. Several open questions remain unanswered, and there is a need to make the methodology available to a broader audience. The aim of this special issue was therefore to stimulate and explore advances in methods, applications, and software for multivariate ANOVA.

The collection of papers includes methodical improvements, practical applications, a tutorial, and a software demonstration. Application areas range from spectroscopic control of fermentation processes to metabolomics and gene expressions. Overall, this issue showcases the power and applicability of multivariate ANOVA methods in a wide range of domains.

I am greatly honored to have been selected as the new Editor-in-Chief of Journal of Chemometrics. I am pleased and enthusiastic to contribute to the history of a journal that has been around since the very early days of chemometrics, and I will do my best to drive it into new developments.

I would like to express my warm and sincere thanks to the outgoing Editor-in-Chief, Age Smilde, for his contribution, leadership and confidence, and my deepest gratitude to Ru Qin Yu who has for many years very actively and served with dedication the Journal as Editor. Building upon the legacy and tradition of my renowned predecessors, and on the significant progress made over the past few years under Age's editorship, I will strive to maintain the excellence of Journal of Chemometrics, attracting impactful papers and disseminating new information in our field.

With Anna de Juan from the University of Barcelona and Hailong Wu from the University of Hunan, the Journal welcomes two new Editors who will complete an engaged team of worldwide international experts serving as members of the editorial and advisory boards. Without losing our identity, we will work to address new trends in data science in chemistry and identify challenging applications of chemometrics. We will continue to expand the space for topical publications, to solicit the best scientific findings through tutorials, perspective papers and feature issues from authors investigating new topics towards novel frontiers, attracting young researches and recognizing their early work. We will also increase our efforts to establish and implement reproducible research and data accessibility. With the help of a dedicated Wiley managing team, the publication cycle will be improved. Authors will continue benefiting from helpful feedback provided by a broad pool of highly competent and dedicated reviewers, who are the pillars of the journal's high quality and reputation. We greatly value their commitment to the journal.

With the aim of enduring the success of Journal of Chemometrics, we will always welcome your comments, suggestions and feedback.

I look forward to interacting with you on a Chemometric occasion.

Edmund R. Malinowski is well known for his factor analysis-based work; he is clearly less well known for his chemical model-based analyses of chemical data. In this contribution, we discuss the, at the time innovative, idea of subjecting the primary model-free analysis results to a secondary quantitative model-based evaluation. Two examples from his research serve as illustrations: the complexation of Cu²⁺ with cyclodextrin and the dimerisation and trimerisation of methylene blue. In both examples, the spectrophotometric titration data are first analysed by window factor analysis, WFA, resulting in the concentration profiles. These primary, model-free results allowed Malinowski to gain qualitative insight in the chemical processes. Additional quantitative analyses of the concentration profiles, based on the previously obtained reaction model, resulted in the numerical values of the underlying equilibrium constants. This contribution relates and compares this methodology with alternative ways of analysing the same data.

Fluorescence spectroscopy combined with parallel factor analysis (PARAFAC) has successfully been applied for the analysis of food and beverages containing numerous autofluorescent compounds. For the decomposition of such data, it is crucial to establish the PARAFAC model complexity. This is not a trivial matter, especially when the sample complexity increases. Diagnostics are available for assisting the choice of the number of PARAFAC components, such as the core consistency. In this short communication, we show that when it comes to real (complex) data, the core consistency is too conservative and other diagnostic tools must be taken into account. We emphasize that it is imperative to inspect the PARAFAC excitation and emission loadings and assess whether these are chemically meaningful.