Network inference and community detection, based on covariance matrices, correlations and test statistics from arbitrary distributions

TL;DR

This paper introduces a methodology to infer binary network adjacency matrices from various association measures, enabling community detection in high-dimensional data with efficient algorithms.

Contribution

It presents a novel approach to derive binary adjacency matrices from diverse association measures, facilitating network analysis from arbitrary distributions.

Findings

Applicable to large high-dimensional datasets

Uses computationally efficient algorithms

Demonstrated utility across various contexts

Abstract

In this paper we propose methodology for inference of binary-valued adjacency matrices from various measures of the strength of association between pairs of network nodes, or more generally pairs of variables. This strength of association can be quantified by sample covariance and correlation matrices, and more generally by test-statistics and hypothesis test p-values from arbitrary distributions. Community detection methods such as block modelling typically require binary-valued adjacency matrices as a starting point. Hence, a main motivation for the methodology we propose is to obtain binary-valued adjacency matrices from such pairwise measures of strength of association between variables. The proposed methodology is applicable to large high-dimensional data-sets and is based on computationally efficient algorithms. We illustrate its utility in a range of contexts and data-sets.