Structured Regression Models

• Applicant: Professor Dr. Walter Zucchini, since 2/2011
• Subject Area: Statistics and Econometrics
• Term: from 2008 to 2012
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 40095828

The project Semiparametric Structured Models is mainly the continuation of the former called structured regression models project. The title modification accounts for the fact that in the end a good part of the project was concerned with densities, distributions or some macro-parameters. Nonetheless, the project continues to comprise the fields Dimension Reduction (via separability - like additivity - or generalized PCA), Semi-mixed effects models (or say, semiparametric mixed effects models), and its Applications, where we will focus on. As before, this project does closely cooperate with [A3] and [A5], but naturally also with the new one of [B5], which strongly overlaps with our interests in semiparametric mixed effects models. We are searching for feasible methods to estimate and analyze model structures which are of particular interest in econometrics for its interpretation (separability and PCAs), and in mathematical statistics for its properties (flexibility and dimensionality). In the next three years these will be even more combined with the particular question of regularization parameter choices. This is still a quite new topic for semi- or nonparametric mixed effects models. Again, a main interest of our project is to bring mathematical developments into practice. This includes also a data orientated modelling and all details coming along with the implementation.

• DFG Programme: Research Units
• Subproject of: FOR 916: Statistical Regularisation and Qualitative Constraints - Inference, Algorithms, Asymptotics and Applications
• Ehemaliger Antragsteller: Professor Dr. Stefan Sperlich, until 1/2011