Latent Variables Models with Simple Structure for Clustering and Data Reduction

Clustering and Data Reduction are relevant topics in modern statistics due to the continuous production of large multivariate data sets. Relevant methodologies are needed to allow easy interpretation of this data deluge. In this paper new latent variable models with simple structure and object clustering are proposed and discussed together with efficient coordinate ascent algorithms. Some examples are given to illustrate the new methodologies and assess their performance.