Mathematics and Statistics Colloquium

  • October 28, 2021
  • 4:30 PM - 5:30 CST
  • MUND 514
  • Tuyen Tran, ttran18@luc.edu
  • Not open to the public.
  • Add to calendar
  • Details

    Speaker: Sameer Deshpande

    Title: Simultaneous variable and covariance selection with the multivariate spike-and-slab LASSO

    Abstract: We consider multivariate linear regression models in which the goal is to predict q possibly correlated responses using a common set of p predictors. In these problems, interest lies not only in determining whether or not a particular predictor has an effect on each response but also in understanding the residual dependence between the outcomes. We propose a Bayesian procedure for such determination using continuous spike-and-slab priors. Rather than relying on a stochastic search through the high-dimensional parameter space, we develop an Expectation-Maximization algorithm targeting modal estimates of the matrix of regression coefficients and residual precision matrix. A key feature of our method is the model of our uncertainty about which parameters are negligible. Essentially, this model enables us to shrink parameters to zero in an automatic and data-adaptive fashion. Our method is seen to substantially outperform regularization competitors that employ fixed penalties on simulated data. We demonstrate our method with a re-examination of data from a recent observational study of the effect of playing high school football on several late-life cognitive psychological, and socioeconomic outcomes.

    About the speaker: Dr. Deshpande is an Assistant Professor in Statistics Department at the University of Wisconsin ¿ Madison. His research interests include Bayesian hierarchical modeling, treed regression, model selection, and causal inference with applications in public health and sports.