Research Seminar
The family of hierarchical classes models: State of the artIven Van MechelenKU Leuven | |
| Abstract: | Hierarchical classes models consitute a distinct family of models for two- or multiway data. The models include an approximate reconstruction of the data, along with clusterings of each of the modes involved in the data and a representation of the structure of (generalized) implication-type relations that can be naturally defined on these modes. A special feature of the models is that they go with a comprehensive graphical representation. In this talk, I will present a state-of-the-art overview of research on the hierarchical classes family. The overview will start on a deterministic level with an introduction into the original (disjunctive) hierarchical classes model for binary two-way data as developed by De Boeck and Rosenberg (1988); this model will further be reconceptualized as a particular type of biclustering model. Subsequently, we will move on to a stochastic level with a discussion of several possible extensions of the deterministic model with distributional assumptions. Then, the estimation of deterministic as well as stochastic hierarchical classes models will be considered, along with issues to be dealt with in the analysis of data on the basis of the models. We conclude with an overview of recent research on several generalizations of the original hierarchical classes model, including (a) conjunctive models, (b) models for rating- and real-valued data, (c) models for multiway data, and (d) models for multiblock data or data fusion. |
| Date: | Tue Nov 13, 12:15 pm - 1:15 pm |
| Place: | room 00.60 (Department of Psychology, Tiensestraat 102, 3000 Leuven) |