Finite mixtures of regressions with random covariates using multivariate skewed distributions

Finite mixtures of regressions with random covariates (CWMs) are a com- mon model-based clustering methodology. Despite a lot of distributions have been considered for both the responses and covariates, skewed distributions have not yet been considered in this framework. Here, a family of 24 novel CWMs is introduced, where both the covariates and response variables are modeled by using one of the four considered skewed distributions, or the Gaussian distribution. A simulated data example is illustrated.