Revistes Publicacions URV: SORT - Statistics and Operations Research Transactions> 2013

New insights into evaluation of regression models through a decomposition of the prediction errors: application to near-infrared spectral data

  • Identification data

    Identifier: RP:2394
    Handle: https://hdl.handle.net/20.500.11797/RP2394
    Authors:
    Urbano, Francisco JoséMarinas, Jose MªMarinas, AlbertoCaridad, José MªSánchez-López, ElenaSánchez-Rodríguez, María Isabel
    Abstract:
    This paper analyzes the goodness of linear regression models taking into account usual criteria such as the number of principal components or latent factors, the goodness of fit or the predictive capability. Other comparison criteria, more common in an economic context, are also considered: the degree of multicollinearity and a decomposition of the mean squared error of the prediction which determines the nature, systematic or random, of the prediction errors. The applications use real data of extra-virgin oil obtained by NIR spectroscopy. The great dimensionality of the data is reduced by applying principal component analysis (PCA) and partial least squares (PLS) analysis. A possible improvement of PCA and PLS regressions by using cluster analysis or the information of the relative maxima of the spectrum is investigated. Finally, obtained results are generalized via cross-validation and bootstrapping

  • Others:

    URV's Author/s: Urbano, Francisco José Marinas, Jose Mª Marinas, Alberto Caridad, José Mª Sánchez-López, Elena Sánchez-Rodríguez, María Isabel
    Keywords: Principal components, partial least squares, multivariate calibration, NIR spectroscopy
    Abstract: This paper analyzes the goodness of linear regression models taking into account usual criteria such as the number of principal components or latent factors, the goodness of fit or the predictive capability. Other comparison criteria, more common in an economic context, are also considered: the degree of multicollinearity and a decomposition of the mean squared error of the prediction which determines the nature, systematic or random, of the prediction errors. The applications use real data of extra-virgin oil obtained by NIR spectroscopy. The great dimensionality of the data is reduced by applying principal component analysis (PCA) and partial least squares (PLS) analysis. A possible improvement of PCA and PLS regressions by using cluster analysis or the information of the relative maxima of the spectrum is investigated. Finally, obtained results are generalized via cross-validation and bootstrapping
    Journal publication year: 2013
    Publication Type: info:eu-repo/semantics/publishedVersion info:eu-repo/semantics/article

  • Keywords:

    Principal components, partial least squares, multivariate calibration, NIR spectroscopy

  • Documents:

    DocumentPrincipal

  • Cerca a google

    Search to google scholar