Seemingly unrelated clusterwise linear regression for contaminated data

Abstract Clusterwise regression is an approach to analysis based on finite mixtures which generally employed when sample observations come from a population composed of several unknown sub-populations. Whenever the response continuous, Gaussian clusterwise linear models are usually employed. Such have been recently robustified with respect possible presence mild outliers in However, some fields research, especially modelling multivariate economic data or social sciences, there may be prior information specific covariates considered term prediction certain response. As consequence, not same for all responses. Thus, novel class proposed. This provides extension mixture-based and correlated responses that let researcher free use different vector each Details about model identification maximum likelihood estimation via expectation-conditional maximisation algorithm given. The performance new studied by simulation comparison other models. A comparative evaluation their effectiveness usefulness provided through real dataset.