Abstract— In contrast to the usual "objects-features" scheme widely used in classification, this paper explores a case where original immeasurable features fall apart into a set of auxiliary features with measurable (in some scales) values. In this case original features are called hierarchical ones. We consider the aggregation procedures for auxiliary features, the application of aggregation results for assigning of classes on the set of unlabeled objects of the sample chosen, and the classification issues with using assigned classes.