In this paper, we propose a novel approach for tackling the obstacles of empirical likelihood in the face of massive data, which is called split sample mean empirical likelihood (SSMEL), our approach provides a unique perspective for solving big data problems. We show that the SSMEL estimator has the same estimation efficiency as the empirical likelihood estimator with the full dataset, and maintains the important statistical property of Wilks' theorem, allowing our proposed approach to be used for statistical inference without estimating the covariance matrix. This effectively tackles the hurdle of the Divide and Conquer (DC) algorithm for statistical inference. We further illustrate the proposed approach via simulation studies and real data analysis.
翻译:暂无翻译