Multivariate Normality Test with Copula Entropy

Jian Ma
DOI: https://doi.org/10.48550/arXiv.2206.05956
2022-06-13
Methodology
Abstract:In this paper, we proposed a multivariate normality test based on copula entropy. The test statistic is defined as the difference between the copula entropies of unknown distribution and the Gaussian distribution with same covariances. The estimator of the test statistic is presented based on the nonparametric estimator of copula entropy. Two simulation experiments were conducted to compare the proposed test with the five existing ones. Experiment results show the advantage of our test over the others.
What problem does this paper attempt to address?