Project Details
Projekt
Sublinear time methods with statistical guarantees
Applicants
Professor Dr. Holger Dette; Professor Dr. Axel Munk
Subject Area
Mathematics
Term
since 2021
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 460867398
In this project, we will develop statistical methods which rely on sub- and resampling strategies andinvestigate their statistical efficiency. Our common idea and fundamental approach is to systematically exploit the fact that in a given sample, observations typically carry a different amount of information and to identify the “most informative data-points”. From this perspective we will reconsider the common linear model and various extensions such as high-dimensional and generalized linear models, or monotone, quantile, logistic and change point regression. In such models, we study inference for the unknown parameters after subsampling from massive data. In our investigations, sequential (i.e. data adaptive) techniques will be fundamental. For example, in change point regression they are used to dynamically determine the next search location given the previous ones, while in generalized linear models they will serve as efficient preliminary estimates of the sampling weights for optimal subsampling.
DFG Programme
Research Units
