Advances in Model-Based Clustering with the Multivariate Contaminated Normal Distribution

Antonio Punzo, PhD
Professor of Statistics, Economics and Business, University of Catania, Italy

The multivariate contaminated normal (MCN) distribution offers a straightforward elliptical heavy-tailed extension of the multivariate normal (MN) distribution, specifically designed to handle and identify mild outliers—often referred to as “bad” points in the MCN literature. Notably, its two additional parameters provide a practical interpretation representing the proportion of good observations and the degree of contamination. In this talk, I will review the applications of the MCN distribution, along with some of its extensions, in mixture model-based clustering.

Register to attend via Zoom