Online Program Home
My Program

Abstract Details

Activity Number: 656
Type: Contributed
Date/Time: Thursday, August 4, 2016 : 8:30 AM to 10:20 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #321031
Title: Testing Statistical Significance of Canonical Correlation Coefficients
Author(s): Yunjin Choi* and Robert Tibshirani and Jonathan Taylor
Companies: Stanford University and Stanford University and Stanford University
Keywords: Canonical Correlation Analysis ; Multivariate data analysis ; Post-selection inference ; Dimension reduction

Given plural datasets, Canonical Correlation Analysis (CCA) investigates the linear transformation of the variables which reduces the correlation structure to the simplest possible form, and addresses the relationships between the variables among the datasets. We propose a novel method for testing the statistical significance of canonical correlation coefficients between two datasets. Utilizing post-selection inference framework, our proposed method provides exact type I error as well as steady detection power with Gaussian assumption. Simulation results compare well with existing approaches.

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

Back to the full JSM 2016 program

Copyright © American Statistical Association