Incorporating Clustering Techniques into GAMLSS

A method for statistical analysis of multimodal and/or highly distorted data is presented. The new methodology combines different clustering methods with the GAMLSS (generalized additive models location, scale, and shape) framework, therefore called c-GAMLSS, “clustering GAMLSS. ” In this extended structure, a latent variable (cluster) created to explain response-variable (target). Any all parameters distribution response can also be modeled by functions covariate added other available resources (features). selecting used carried out in stages, step-based method. simulation study considering multiple scenarios presented compare c-GAMLSS existing Gaussian mixture models. We show means four applications that cases where authentic explanatory variables are or not available, structure outperforms models, some recently developed complex distributions, cluster-weighted mixture-of-experts model. Even though we use simple distributions our examples, more sophisticated variable.