Аннотация:When building classification models of complex systems with many classes, the traditionalchemometric approaches such as discriminant analysis or soft independent modelling of classanalogy often fail. Some people resort to advanced deep neural network, but this is only anoption if there is access to very many samples. Another alternative often used is to buildhierarchical models where subclasses are sort of peeled off one or a few at a time. Suchapproaches often outperform classical classification as well as deep neural network on smallmulti-class problems. The downside though, is that it is very cumbersome to build suchhierarchies of models. It requires substantial work of a skilled person. In this paper, wedevelop a fully automated approach for building hierarchical models and test the performanceon a number of classification problems.