Title: On the compositional interpretation of frequency data
Authors: Tamas Rudas - Hungarian Academy of Sciences Centre for Social Sciences (Hungary) [presenting]
Anna Klimova - National Center for Tumor Diseases Partner Site Dresden TU Dresden (Germany)
Abstract: The aim is to discuss the feasibility of the standard normalization step in the compositional interpretation of frequency data, relying on some recent results in categorical data analysis. In the latter setup, the target of the statistical inference is the probability distribution underlying the data. Recent results indicate that depending on the model one has for the underlying probability distribution, the normalization may not be feasible. In the case of compositional data, the assumed distribution for the compositions is parallel to the model for the probability distribution in the case of categorical data analysis. In cases, when normalization is not possible, a proper analysis will be suggested and illustrated with data from cell biology.