Online Program Home
My Program

Abstract Details

Activity Number: 253 - Contributed Poster Presentations: Section on Statistical Computing
Type: Contributed
Date/Time: Monday, July 30, 2018 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistical Computing
Abstract #328869
Title: Feature Selection and Classification Using Sparse Envelope Model
Author(s): Minji Lee* and Zhihua Su
Companies: University of Florida and University of Florida
Keywords: Feature selection; Dimension reduction; Envelope model; Multi-class classification

We propose a new method for multi-class classification and feature selection using the sparse envelope model. The sparse envelope model (Su et al., 2016) can conduct variable selection on the responses in a multivariate regression model and achieve the efficiency gains compared to the standard model. We apply the sparse envelope model to a one-way multivariate analysis of variance which enable it to perform a feature selection in this context. We found that, even though the part of features is not significant, non-selected features should not be removed to improve efficiency of significant features. Simulation studies show that our method has a selection consistency and lower misclassification rates than some recent methods. Consistency and the oracle property of the proposed model are established and the asymptotic distribution of the estimator is obtained.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2018 program