Networks consist of interconnected units, known as nodes, and allow to formally describe interactions within a system. Specifically, bipartite networks depict relationships between two distinct sets of nodes, designated as sending and receiving nodes. An integral aspect of bipartite network analysis often involves identifying clusters of nodes with similar behaviors. The computational complexity of models for large bipartite networks poses a challenge. To mitigate this challenge, we employ a Mixture of Latent Trait Analyzers (MLTA) for node clustering. Our approach extends the MLTA to include covariates and introduces a double EM algorithm for estimation. Applying our method to COVID-19 data, with sending nodes representing patients and receiving nodes representing preventive measures, enables dimensionality reduction and the identification of meaningful groups. We present simulation results demonstrating the accuracy of the proposed method.
25 November 2024 in Statistical Methods & Applications
In the context of network data, bipartite networks are of particular interest, as they provide a useful description of systems representing relationships between sending and receiving nodes. In this framework, we extend the mixture of latent trait analyzers (MLTA) model with concomitant variables (nodal attributes) to perform a joint clustering of the two disjoint sets of nodes of a bipartite network, as in the biclustering framework. In detail, sending nodes are partitioned into clusters (called components) via a finite mixture of latent trait models. In each component, receiving nodes are partitioned into clusters (called segments) by adopting a flexible and parsimonious specification of the linear predictor. Residual dependence between receiving nodes is modeled via a multidimensional latent trait, as in the original MLTA specification. Furthermore, by incorporating nodal attributes into the model’s latent layer, we gain insight into how these attributes impact the formation of components. To estimate model parameters, an EM-type algorithm based on a Gauss-Hermite approximation of intractable integrals is proposed. A simulation study is conducted to test the performance of the model in terms of clustering and parameters’ recovery. The proposed model is applied to a bipartite network on pediatric patients possibly affected by appendicitis with the objective of identifying groups of patients (sending nodes) being similar with respect to subsets of clinical conditions (receiving nodes).
Analyzing three-way data is challenging due to complex dependencies between observations, which must be accounted for to ensure reliable results. We focus on hierarchical, multivariate, binary data organized in a three-way data structure, where rows correspond to first-level units, columns to variables, and layers to second-level units within which the first-level units are nested. In this framework, model-based clustering methods can be effectively employed for dimensionality reduction purposes, facilitating a clear understanding of the phenomenon under investigation. In this work, we propose a novel modeling tool for a hierarchical clustering of first- and second-level units. We extend the Mixture of Latent Trait Analyzers (MLTA) with concomitant variables by letting prior component probabilities depend also on second-level-specific random effects. Parameter estimation is performed by means of a double EM algorithm based on a variational approximation of the model log-likelihood function, along with a nonparametric maximum likelihood estimation of the second-level-specific random effect distribution. This latter approach allows to estimate a discrete distribution which directly provides a clustering of second-level units. Within (conditional on) each of such clusters, first-level units are partitioned thanks to the MLTA specification. The proposal is applied to data from the European Social Survey to partition countries (second-level units) according to the baseline attitude of their residents (first-level units) toward digital technologies (variables). Within these clusters, residents are partitioned on the basis of their attitude toward specific digital skills. The influence of socio-economic factors on the identification of digitalization profiles is also taken into consideration via a concomitant variable approach.