Mixture decomposition of distributions by copulas in the symbolic data analysis framework

摘要: This work investigates the situation in which each unit from a given set is described by some vector of p probability distributions. Our aim to find simultaneously ''good'' partition these units and probabilistic description clusters with model using ''copula functions'' associated class this partition. Different copula models are presented. The mixture decomposition problem resolved general case. result extends standard case where distributions instead traditional classical single (categorical or numerical) values. Several generalizations algorithms proposed. All results first considered variable then extended variables top-down binary tree approach.