Simultaneous Dimension Reduction and Variable Selection in Multivariate Regression

• Activity Number: 568
• Type: Contributed
• Date/Time: Wednesday, August 1, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Learning and Data Mining
• Author(s): Lisha Chen*+ and Jianhua Huang
• Companies: Yale University and Texas A&M University
• Keywords: dimesion reduction ; variable selection ; multivariate regression

Abstract

In this talk, we will introduce a new method to address the problem of predicting several response variables from the same set of predictor variables using linear regression. The method incorporates the interrelation between the response variables to improve the overall predictive accuracy. When the dimension of the predictors is high, the new proposal conducts variable selection and dimension reduction simultaneously. We will discuss the asymptotic consistency of the proposed method. The new procedure is compared with several previously proposed variable selection methods for multivariate regression and exhibits improved accuracy in prediction and variable selection.