Bayesian Markov Switching Tensor Regression for Time-varying Networks

TL;DR

This paper introduces a Bayesian Markov switching tensor regression model for analyzing time-varying multilayer networks, incorporating low-rank tensor decomposition and hidden Markov chains for structural change detection.

Contribution

It presents a novel Bayesian framework with low-rank tensor parametrization and Markov switching dynamics for modeling complex time-varying network data.

Findings

Effective modeling of financial network dynamics.

Identification of structural changes in network topology.

Demonstrated improved inference with Gibbs sampling.

Abstract

We propose a new Bayesian Markov switching regression model for multidimensional arrays (tensors) of binary time series. We assume a zero-inflated logit regression with time-varying parameters and apply it to multilayer temporal networks. The original contribution is threefold. First, to avoid over-fitting we propose a parsimonious parametrization based on a low-rank decomposition of the tensor of regression coefficients. Second, we assume the parameters are driven by a hidden Markov chain, thus allowing for structural changes in the network topology. We follow a Bayesian approach to inference and provide an efficient Gibbs sampler for posterior approximation. We apply the methodology to a real dataset of financial networks to study the impact of several risk factors on the edge probability. Supplementary materials for this article are available online.