Title: Two novel resampling strategies for dependent data
Authors: Srijan Sengupta - North Carolina State University (United States) [presenting]
Xiaofeng Shao - University of Illinois at Urbana-Champaign (United States)
Stanislav Volgushev - University of Toronto (Canada)
Abstract: Two novel resampling strategies are proposed, namely, the dependent random weighting (DRW) and the subsampled double bootstrap (SDB). The DRW is a generalization of the traditional random weighting where the weights are made to be temporally or spatially dependent and are adaptive to the configuration of the data. Unlike the block-based bootstrap or subsampling methods, the DRW can be used for irregularly spaced time series and spatial data without any implementational difficulty. The SDB is a fast resampling strategy for massive data applications which are increasingly prevalent. For massive datasets, classical bootstrap strategies (and its block-based versions) becomes prohibitively costly in computation even with modern parallel computing platforms. Recently a method called BLB (Bag of Little Bootstraps) for massive data has been proposed which is more computationally scalable with little sacrifice of statistical accuracy. Building on BLB and the idea of fast double bootstrap, we propose the SDB for both independent data and time series data. For both new methods, we establish theoretical properties and demonstrate empirical performance in simulation studies and data analysis.