Title: Clustering multivariate count data via a mixture of Poisson factor models
Authors: Yang Tang - McMaster University (Canada) [presenting]
Paul McNicholas - McMaster University (Canada)
Abstract: Dependencies in multivariate counts are of interest in many applications, but few approaches have been proposed for their analysis. We develop a mixture of Poisson factor models to explore the dependencies in multivariate count data. We assume that the $p$-dimensional random vector of counts $Y$ is modeled using a $q$-dimensional vector of latent factors $U$ and $U$ follows an inverse Gaussian distribution. The proposed framework provides a parsimonious and easy to interpret representation of multivariate dependencies in counts. Parameter estimation is carried out and information criteria are used for model selection. The use of these models is demonstrated on real and simulated data.