Open Conference Systems, CLADAG2023

Font Size: 
Model-based clustering for torus data
Luca Greco, Antonio Lucadamo, Claudio Agostinelli

Last modified: 2023-07-07


Torus data are defined as multivariate circular observations that arise as measurements on a periodic scale and are recorded as angles. In this paper, by paralleling a methodology for Euclidean data, we focus on parsimonious model based clustering for torus data by building on the mclust methodology. Therefore, covariance constraints are imposed on the completely general heterogeneous clustering model allowing a flexible and general framework to clustering torus data.