Jove
Visualize
Contact Us

Related Experiment Videos

Spatial sign preprocessing: a simple way to impart moderate robustness to multivariate estimators.

Sven Serneels1, Evert De Nolf, Pierre J Van Espen

  • 1Department of Chemistry, University of Antwerp, Universiteitsplein 1, 2610 Antwerpen, Belgium. sven.serneels@ua.ac.be

Journal of Chemical Information and Modeling
|May 23, 2006
PubMed
Summary

Spatial sign transformation offers robust preprocessing for multivariate data analysis. While efficient, it introduces bias, impacting partial least squares regression performance in chemometrics and QSAR studies.

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by
Same author

How to construct a multiple regression model for data with missing elements and outlying objects.

Analytica chimica actaยท2007
See all related articles
JoVE
x logofacebook logolinkedin logoyoutube logo
ABOUT JoVE
OverviewLeadershipBlogJoVE Help Center
AUTHORS
Publishing ProcessEditorial BoardScope & PoliciesPeer ReviewFAQSubmit
LIBRARIANS
TestimonialsSubscriptionsAccessResourcesLibrary Advisory BoardFAQ
RESEARCH
JoVE JournalMethods CollectionsJoVE Encyclopedia of ExperimentsArchive
EDUCATION
JoVE CoreJoVE BusinessJoVE Science EducationJoVE Lab ManualFaculty Resource CenterFaculty Site
Terms & Conditions of Use
Privacy Policy
Policies

Area of Science:

  • Multivariate statistics
  • Chemometrics
  • Robust data analysis

Background:

  • Spatial sign provides a robust multivariate extension of the sign concept.
  • Multivariate covariance estimators using spatial signs are computationally efficient and robust to outliers.
  • Spatial sign transformation can serve as a robust preprocessing technique.

Purpose of the Study:

  • To investigate the efficacy of spatial sign transformation as a preprocessing step for partial least squares regression (PLS).
  • To evaluate the performance of spatial sign-PLs in comparison to non-transformed data and robust PLs methods.
  • To assess the application of spatial sign-PLs in quantitative structure-activity relationship (QSAR) analysis.

Main Methods:

  • Simulation study comparing spatial sign transformation with non-transformed data and robust PLs methods.

Related Experiment Videos

  • Application of spatial sign-PLs to a QSAR dataset.
  • Evaluation of robustness and efficiency of the spatial sign transformation.

    • Main Results:

    • Spatial sign transformation is computationally simple and robust to outliers.
    • The spatial sign transform demonstrates fair efficiency but exhibits undesirable bias properties.
    • Spatial sign-PLs performed comparably to the best linear model on a QSAR dataset.

    Conclusions:

    • Spatial sign transformation is a viable, robust preprocessing technique for PLs, particularly in chemometrics.
    • The observed bias in spatial sign transformation warrants careful consideration in specific applications.
    • The method shows promise for QSAR studies, offering competitive performance with existing models.