Journal of Chemometrics

Cover image for Vol. 27 Issue 3-4

March-April 2013

Volume 27, Issue 3-4

Pages i–iii, 43–90

  1. Issue Information

    1. Top of page
    2. Issue Information
    3. Research Articles
    1. Issue Information (pages i–iii)

      Version of Record online: 10 APR 2013 | DOI: 10.1002/cem.2495

  2. Research Articles

    1. Top of page
    2. Issue Information
    3. Research Articles
    1. A novel tree kernel partial least squares for modeling the structure–activity relationship (pages 43–49)

      Xin Huang, Dong-Sheng Cao, Qing-Song Xu and Yi-Zeng Liang

      Version of Record online: 3 MAR 2013 | DOI: 10.1002/cem.2490

      Tree kernel partial least squares classification algorithm has been developed by constructing an informative kernel on the basis of decision tree ensemble, which can select some informative features by variable importance ranking and deal with the nonlinear classification problems.

    2. Sparse models by iteratively reweighted feature scaling: a framework for wavelength and sample selection (pages 50–62)

      Erik Andries

      Version of Record online: 20 FEB 2013 | DOI: 10.1002/cem.2492

      Recently, there has been an increase in the use of sparse multivariate calibration methods in chemometrics. However, sparse methods are still not as well understood or as fast as their classical counterparts (such as partial least squares). In this paper, we describe a simple framework whereby classical multivariate calibration methods can be iteratively used to generate sparse models. Moreover, this approach allows for either wavelength or sample sparsity.

    3. Solving matrix effect, spectral overlapping and nonlinearity by generalized standard addition method coupled with radial basis functions–partial least squares: simultaneous determination of atorvastatin and amlodipine in urine (pages 63–69)

      Masoud Shariati-Rad, Mohsen Irandoust, Tayyebeh Amini and Mojtaba Shamsipur

      Version of Record online: 20 FEB 2013 | DOI: 10.1002/cem.2491

      Nonlinearity in multivariate data was identified by principal component analysis. Generalized standard addition data were processed by radial basis functions–partial least squares for solving matrix effects, spectral interferences and nonlinearity in multivariate calibration.

    4. Solving the sign indeterminacy for multiway models (pages 70–75)

      Rasmus Bro, Riccardo Leardi and Lea Giørtz Johnsen

      Version of Record online: 18 MAR 2013 | DOI: 10.1002/cem.2493

      Bilinear and multilinear models such as principal component analysis and PARAFAC have intrinsic sign indeterminacies. For example, any loading vector can be multiplied by −1 if another vector of that particular component is also multiplied by −1 without affecting the loss function values. This sometimes causes problems, for example, with respect to interpretation. In this paper, a method is developed to fix the sign indeterminacy for the PARAFAC, Tucker3 and PARAFAC2 models.

    5. Bridging the gap between metabolic profile determination and visualization in neurometabolic disorders: a multivariate analysis of proton magnetic resonance in vivo spectra (pages 76–90)

      Agnieszka Skorupa, Ewa Jamroz, Justyna Paprocka, Maria Sokół, Magdalena Wicher and Aleksandra Kiełtyka

      Version of Record online: 21 MAR 2013 | DOI: 10.1002/cem.2494

      Principal component analysis was coupled with proton magnetic resonance spectroscopy. The overlap between neurometabolic and other neurological disorders has to be checked. Separation of various groups of spectra was analyzed after dimensionality reduction. Several neurometabolic disorders (metachromatic leukodystrophy, globoid leukodystrophy, Canavan disease, megalencephalic leukoencephalopathy with subcortical cysts) were separated from the main bulk of data. The presented technique is an efficient tool in exploration of heterogeneous datasets.