A0542
Title: Optimal subsampling for hierarchical data
Authors: Songqiao Han - Kings College London (United Kingdom) [presenting]
Kalliopi Mylona - King's College London (United Kingdom)
Steven Gilmour - KCL (United Kingdom)
Abstract: Hierarchical data analysis is an important topic in big data research. However, the computational costs associated with parameter estimation and model fitting in large datasets are very high, making efficient subsampling techniques necessary. An optimal subsampling method is introduced specifically designed for hierarchical data, which fully considers the connections between and within different levels. This method enables optimal subsampling across various scenarios. In addition, several examples of industrial data applications are given.
