Testing the equality of two covariance matrices for non-normal high-dimensional data
Keywords:
high-dimensional covariance matrices, hypothesis test, equality, non-normalityAbstract
This paper focuses on testing the equality of two high-dimensional covariance matrices without the normality assumption. Two new test statistics are proposed by taking the ratio between the unbiased and consistent estimators of the trace of the covariance matrices. Under some mild assumptions, the proposed test statistics are proved to be asymptotically normal. Furthermore, numerical simulations demonstrate that the proposed tests have good size and power with varying dimensions and sample sizes.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2020 Jieqiong Shen

This work is licensed under a Creative Commons Attribution 4.0 International License.
L'opera è pubblicata sotto Licenza Creative Commons Attribuzione 4.0 Internazionale (CC-BY)

