Model-based clustering for noisy longitudinal circular data, with application to animal movement

In this work, we introduce a model for circular data analysis to robustly estimate parameters, under a longitudinal clustering setting. A hidden Markov model for longitudinal circular data combined with a uniform conditional density on the circle to capture noise observations is proposed. A set of exogenous covariates is available; they are assumed to affect the evolution of clustering over time. Parameter estimation is carried out through a hybrid expectation–maximization algorithm, using recursions widely adopted in the hidden Markov model literature. Examples of application of the proposal on real and simulated data are performed to show the effectiveness of the proposal.