Skip NavigationSkip to Content

An efficient stochastic search for Bayesian variable selection with high-dimensional correlated predictors

  1. Author:
    Kwon, D.
    Landi, M. T.
    Vannucci, M.
    Issaq, H. J.
    Prieto, D.
    Pfeiffer, R. M.
  2. Author Address

    [Kwon, D; Landi, MT; Pfeiffer, RM] NCI, Div Canc Epidemiol & Genet, Rockville, MD 20852 USA [Vannucci, M] Rice Univ, Dept Stat, Houston, TX 77251 USA [Issaq, HJ; Prieto, D] SAIC Frederick Inc, Lab Prote & Analyt Technol, Frederick, MD 21702 USA;Kwon, D (reprint author), NCI, Div Canc Epidemiol & Genet, Rockville, MD 20852 USA;kwonde@mail.nih.gov
    1. Year: 2011
    2. Date: Oct
  1. Journal: Computational Statistics & Data Analysis
    1. 55
    2. 10
    3. Pages: 2807-2818
  2. Type of Article: Article
  3. ISSN: 0167-9473
  1. Abstract:

    We present a Bayesian variable selection method for the setting in Which the number of independent variables or predictors in a particular dataset is much larger than the available sample size. While most of the existing methods allow some degree of correlations among predictors but do not consider these correlations for variable selection, our method accounts for correlations among the predictors in variable selection. Our correlation-based stochastic search (CBS) method, the hybrid-CBS algorithm, extends a popular search algorithm for high-dimensional data, the stochastic search variable selection (SSVS) method. Similar to SSVS, we search the space of all possible models using variable addition, deletion or swap moves. However, our moves through the model space are designed to accommodate correlations among the variables. We describe our approach for continuous, binary, ordinal, and count outcome data. The impact of choices of prior distributions and hyperparameters is assessed in simulation studies. We also examined the performance of variable selection and prediction as the correlation structure of the predictors varies. We found that the hybrid-CBS resulted in lower prediction errors and identified better the true outcome associated predictors than SSVS when predictors were moderately to highly correlated. We illustrate the method on data from a proteomic profiling study of melanoma, a type of skin cancer. Published by Elsevier B.V.

    See More

External Sources

  1. DOI: 10.1016/j.csda.2011.04.019
  2. WOS: 000292662700003

Library Notes

  1. Fiscal Year: FY2011-2012
NCI at Frederick

You are leaving a government website.

This external link provides additional information that is consistent with the intended purpose of this site. The government cannot attest to the accuracy of a non-federal site.

Linking to a non-federal site does not constitute an endorsement by this institution or any of its employees of the sponsors or the information and products presented on the site. You will be subject to the destination site's privacy policy when you follow the link.

ContinueCancel