关键词: 自助法, 不确定性, 大规模数据, 并行计算

Abstract: The large volume of massive dataset is the key feature of Big Data which brings the main challengers for computing. The dataset with large sample size cannot always stands for the population, therefore the algorithms design for Big Data should consider how to reduce computing cost and how to characterize the uncertainty of the estimated results. In this paper, we study the design of statistical algorithms for massive dataset considering both computing efficiency and uncertainty assessment. Both the Bag of Little Bootstrap (BLB) and Subsampled Double Bootstrap (SDB) algorithms are discussed as illustrative examples. Additionally, a comparison of BLB and SDB is discussed with conclusions of future work.

Key words: Bootstrap, Uncertainty, Massive Data, Parallel Computing