In past few decades latent variable models have become mainstream models when analysing complex, multivariate data. Such data are collected in various fields of applied science. A prime example is community ecology, where observations of multiple interacting species are collected from a set of samples. The data may include covariates related to study sites and species themselves. Most often temporal and/or spatial correlation in responses is also encountered. In this project, we propose new and innovative latent variable models for the analysis of modern, complex abundance data. We consider theoretical properties of the developed methods, provide fast estimation algorithms for model fitting and implement methods to free statistical software. The methods are illustrated by applying them to datasets from community ecology, but besides ecology, the methods are applicable in other fields of science as well. Health sciences, social sciences, psychology and bioinformatics serve as examples.

2023

2027