A Multi-Way Correlation Coefficient
Methodology
2020-03-06 v1
Abstract
Pearson's correlation is an important summary measure of the amount of dependence between two variables. It is natural to want to generalise the concept of correlation as a single number that measures the inter-relatedness of three or more variables e.g. how `correlated' are a collection of variables in which non are specifically to be treated as an `outcome'? In this short article, we introduce such a measure, and show that it reduces to the modulus of Pearson's in the two dimensional case.
Cite
@article{arxiv.2003.02561,
title = {A Multi-Way Correlation Coefficient},
author = {Benjamin M. Taylor},
journal= {arXiv preprint arXiv:2003.02561},
year = {2020}
}
Comments
10 pages, 5 figures