Pearson's correlation coefficient in the theory of networks: A comment

Abstract

In statistics, the Pearson correlation coefficient $r_{x,y}$ determines the degree of linear correlation between two variables and it is known that $-1 \le r_{x,y} \le 1$. In the theory of networks, a curious expression proposed in [PRL {\bf 89} 208701 (2002)] for degree-degree correlation coefficient $r_{j_i,k_i}, i\in [1,M]$ has been in use. We realize that the suggested form is the conventional Pearson's coefficient for $\{(j_i,k_i), (k_i,j_i)\}$ for $2M$ data points and hence it is rightly dedicated to undirected networks.