Journal of Chemometrics

Cover image for Vol. 30 Issue 9

Early View (Online Version of Record published before inclusion in an issue)

Edited By: Prof. Paul J Gemperline

Impact Factor: 1.873

ISI Journal Citation Reports © Ranking: 2015: 15/123 (Statistics & Probability); 18/56 (Instruments & Instrumentation); 22/59 (Automation & Control Systems); 22/101 (Mathematics Interdisciplinary Applications); 41/75 (Chemistry Analytical); 42/130 (Computer Science Artificial Intelligence)

Online ISSN: 1099-128X


  1. 1 - 9

    1. Multivariate resolution of flattened absorption spectra with weighted least squares; weight selection and rotational ambiguity

      Alexey N. Skvortsov

      Version of Record online: 27 SEP 2016 | DOI: 10.1002/cem.2831

      The current achievements in the areas of multivariate curve resolution (MCR) and weighted least squares methods (WLS) were applied to the analysis of the absorption spectra that were strongly affected by measurement uncertainties (peak flattening/saturation). The advantages and disadvantages of different weighting schemes, and their influence on the ability of WLS-based MCR to recover spectral and concentration profiles and correctly estimate rotational ambiguity were demonstrated. Approaches for migration of MCR methods to WLS were also addressed.


    1. On generalized Borgen plots II: The line-moving algorithm and its numerical implementation

      Annekathrin Jürß, Mathias Sawall and Klaus Neymeyr

      Version of Record online: 21 SEP 2016 | DOI: 10.1002/cem.2815

      Borgen plots are geometric constructions that represent the set of all nonnegative factorizations of spectral data matrices for 3-component systems. These constructions are limited to nonnegative data and result in nonnegative factorizations. Generalized Borgen plots allow factors with small negative entries and can be applied to experimental and noisy spectral data sets. The paper presents and justifies the line-moving algorithm for the geometric construction of generalized Borgen plots. A software implementation in the FACPACK software is introduced.


    1. Penalized eigendecompositions: motivations from domain adaptation for calibration transfer

      Erik Andries

      Version of Record online: 14 SEP 2016 | DOI: 10.1002/cem.2818

      In disciplines outside of chemometrics, particularly computer vision, problems involving calibration updating, e.g., calibration transfer or maintenance, are more recent phenomena, and these problems often go under the umbrella term domain adaptation (DA). We report on penalty-based eigendecompositions, a class of DA methods that has its motivational roots in linear discriminant analysis. These DA methods are coopted for chemometrics-based purposes and are then compared with known calibration updating methods.


    1. Targeting programmed cell death 4 (PDCD4) with biogenic compounds in ARDS by Gaussian process-based QSAR virtual screening

      Sheng-Yun Wang, Jun-Ying Liu, Lv Wang, Li-Wei Duan, Qi-Tong Chen, Jin-Hao Zheng, Zhao-Fen Lin and Wen-Fang Li

      Version of Record online: 14 SEP 2016 | DOI: 10.1002/cem.2827

      Potent PDCD4 mediators are identified from biogenic compounds using a combined strategy of statistical screening and binding assay. Structural analysis reveals that nonbonded interactions confer stability and specificity to the PDCD4-mediator recognition.

    2. Radial basis function neural networks based on projection pursuit approach and solvatochromic descriptors: single and full column prediction of gas chromatography retention behavior of polychlorinated biphenyls

      Zeinabe Hassanzadeh, Parastoo Ebrahimi, Mohsen Kompany-Zareh and Raouf Ghavami

      Version of Record online: 4 SEP 2016 | DOI: 10.1002/cem.2822

      The present article compares the ability of projection pursuit (PP) and principal component analysis (PCA) in dimension reduction. The scores of PP and PCA, by a different number of factors, were used as inputs of radial basis function neural network. Radial basis function was used as a nonlinear regression method in a quantitative structure-retention relationship study of 209 polychlorinated biphenyls. The results demonstrate that the dimension reduction ability of the PP is better than that of the PCA.

    3. On estimation of bias field in MRI images: polynomial vs Gaussian surface fitting method

      Sayan Kahali, Sudip Kumar Adhikari and Jamuna Kanta Sing

      Version of Record online: 31 AUG 2016 | DOI: 10.1002/cem.2825

      This paper presents a comprehensive comparative study between polynomial and Gaussian surface fitting methods for bias field estimation and correction in MRI images. The qualitative and quantitative investigations of the simulation results both on simulated and real-patient MRI images show that the Gaussian surface fitting method yields better results.


    1. You have free access to this content
      Orthogonality, uncorrelatedness, and linear independence of vectors

      Richard G. Brereton

      Version of Record online: 23 AUG 2016 | DOI: 10.1002/cem.2816

      The relationship between linear independence, orthogonality, and uncorrelatedness of vectors is described.


    1. A bootstrap-based method for optimal design of experiments

      O. Paquet-Durand, V. Zettel and B. Hitzmann

      Version of Record online: 10 AUG 2016 | DOI: 10.1002/cem.2820

      In this contribution, an optimal design of experiments for the determination of the parameters of the Peleg model has been performed. For the parameter error estimation, a bootstrap-based approach has been used and has been compared to the normal Cramér-Rao lower bound method. Although bootstrapping is computationally more demanding, it has no requirements on the distributions of the measurements or the parameter values. Therefore, it is more flexible and has the potential to be more accurate.

    2. Compression strategies for the chemometric analysis of mass spectrometry imaging data

      Carmen Bedia, Romà Tauler and Joaquim Jaumot

      Version of Record online: 9 AUG 2016 | DOI: 10.1002/cem.2821

      Application of chemometric methods to high-resolution mass spectrometry images requires a preliminary compression of experimental data. An algorithm for the detection of relevant variables for the analysis of images obtained by mass spectrometry is presented. Evaluation of different approaches is performed considering achieved compression rate, mass spectrometry information loss, and computational resources needed.


  1. 1 - 9