Journal of Chemometrics

Cover image for Vol. 25 Issue 6

June 2011

Volume 25, Issue 6

Pages 287–348

  1. Research Articles

    1. Top of page
    2. Research Articles
    1. Estimating the reversing and non-reversing heat flow from standard DSC curves in the glass transition region (pages 287–294)

      Ramón Artiaga, Jorge López-Beceiro, Javier Tarrío-Saavedra, Carlos Gracia-Fernández, Salvador Naya and José L. Mier

      Article first published online: 30 MAR 2011 | DOI: 10.1002/cem.1347

      A model consisting of a mixture of functions representing the heat capacity of the glassy and non-glassy fractions, the glass transition progress and the enthalpic relaxation is proposed to explain standard DSC curves in the glass transition region. Optimal fittings of the model allowed to estimate the reversing and non-reversing signals from standard DSC data. The estimates resulted to be similar to the signals obtained by MTDSC, but with the advantage of being not affected by the modulation frequency.

    2. Assessing the coefficient of variations of chemical data using bootstrap method (pages 295–300)

      Saeid Amiri and Silvelyn Zwanzig

      Article first published online: 17 MAR 2011 | DOI: 10.1002/cem.1350

      Coefficient of variation is a widely used measure of relative variability of data that can be useful to study the uncertainty in chemical experiment. This work uses the bootstrap method to compare the coefficient of variations.

    3. Model choice and squared prediction errors in PLS regression (pages 301–312)

      Rolf Ergon, Maths Halstensen and Kim H. Esbensen

      Article first published online: 16 APR 2011 | DOI: 10.1002/cem.1356

      Fault detection by use of squared prediction errors (SPE) in X should be based on the identical residuals in the non-orthogonalized PLSR, Bidiag2 and RE-PLSR models, and not from the residuals in the conventional orthogonalized PLSR model. The reason is that faults cannot be expected to be governed by the same covariance structure as normal data.

    4. Development of a highly specific ensemble of topological models for early identification of P-glycoprotein substrates (pages 313–322)

      Mauricio Di Ianni, Alan Talevi, Eduardo A. Castro and Luis E. Bruno-Blanch

      Article first published online: 17 FEB 2011 | DOI: 10.1002/cem.1376

      P-glycoprotein is a promiscuous ATP-dependent transporter responsible for multidrug resistance issues in a wide range of diseases. Here we present a highly specific ensemble of topological models obtained through data fusion, aimed to early recognition of Pgp substrates. The models presented here may be applied in virtual screening campaigns in order to discard potential Pgp substrates at the early stage of drug development projects.

    5. Determination of rank by augmentation (DRAUG) (pages 323–328)

      Edmund R. Malinowski

      Article first published online: 30 MAR 2011 | DOI: 10.1002/cem.1377

      The primary rank of a data matrix can be determined by comparing the residual variances obtained from principal component analysis of the data matrix to those obtained from an augmentated matrix. The augmented matrix is generated by adding a unit rank matrix to the original data matrix. The ratio of the residual variances between adjacent factor levels represents a Fisher ratio that can be used to determine the primary rank of the original data matrix.

    6. The log-bimodal-skew-normal model. A geochemical application (pages 329–332)

      Heleno Bolfarine, Héctor W. Gómez and Luisa I. Rivas

      Article first published online: 17 FEB 2011 | DOI: 10.1002/cem.1378

      The paper introduces a new log-skew-bimodal model which seems to present good performance when applied to substance concentration. Model properties such as moments were investigated. As special cases of the model we have the log-skew-normal and log-normal distributions. Maximum likelihood estimators can be computed by using existing statistical software. A real data application shows that the model has great potential for applied work and is able to successfully replace existing log-normal models.

    7. Quantitative structure–property relationship for surface tension of some common alcohols (pages 333–339)

      Aboozar Khajeh and Hamid Modarress

      Article first published online: 10 FEB 2011 | DOI: 10.1002/cem.1379

      New QSPR models is suggested for predicting the surface tension of alcohols from a large number of descriptors belonging to several different classes by using two linear and non-linear chemometric methods, Genetic function approximation (GFA) and adaptive neuro-fuzzy inference system (ANFIS).

    8. Kohonen classification applying ‘missing variables’ criterion to evaluate the p-boronophenylalanine human-body-concentration decreasing profile of boron neutron capture therapy patients (pages 340–348)

      Jorge Magallanes, Alejandro García-Reiriz, Sara Líberman and Jure Zupan

      Article first published online: 17 MAR 2011 | DOI: 10.1002/cem.1383

      The neutron irradiation dose of a Boron Neutron Capture Therapy patient depends on its boron concentration in blood. This concentration cannot be measured while irradiation is being carried out. To predict such blind period, a model is presented based on Kohonen artificial neural networks. The advantages of this model are: a) it can accept as inputs, vectors with incomplete information and b) their prediction ability and robustness can be automatically improved by feeding it with an increasing number of data.