Optimal Distributed Subsampling for Maximum Quasi-Likelihood Estimators with Massive Data

Nonuniform subsampling methods are effective to reduce computational burden and maintain estimation efficiency for massive data. Existing methods mostly focus on subsampling with replacement due to its high computational efficiency. If the data volume is so large that nonuniform subsampling probabilities cannot be calculated all at once, then subsampling with replacement is infeasible to implement. This paper solves this problem using Poisson subsampling. We first derive optimal Poisson subsampling probabilities in the context of quasi-likelihood estimation under the A- and L-optimality criteria. For a practically implementable algorithm with approximated optimal subsampling probabilities, we establish the consistency and asymptotic normality of the resultant estimators. To deal with the situation that the full data are stored in different blocks or at multiple locations, we develop a distributed subsampling framework, in which statistics are computed simultaneously on smaller partitions of the full data. Asymptotic properties of the resultant aggregated estimator are investigated. We illustrate and evaluate the proposed strategies through numerical experiments on simulated and real data sets.

[1]  R Core Team,et al.  R: A language and environment for statistical computing. , 2014 .

[2]  Runze Li,et al.  Statistical inference in massive data sets , 2012 .

[3]  Charles R. Johnson,et al.  Matrix analysis , 1985, Statistical Inference for Engineers and Data Scientists.

[4]  Jihnhee Yu,et al.  Measures, Integrals and Martingales , 2007, Technometrics.

[5]  Jing Wu,et al.  Online Updating of Statistical Inference in the Big Data Setting , 2015, Technometrics.

[6]  S. Muthukrishnan,et al.  Faster least squares approximation , 2007, Numerische Mathematik.

[7]  Eric R. Ziegel,et al.  Generalized Linear Models , 2002, Technometrics.

[8]  Ping Ma,et al.  A statistical perspective on algorithmic leveraging , 2013, J. Mach. Learn. Res..

[9]  A note on the uniqueness of the quasi-likelihood estimator , 1998 .

[10]  Martin J. Wainwright,et al.  Dual Averaging for Distributed Optimization: Convergence Analysis and Network Scaling , 2010, IEEE Transactions on Automatic Control.

[11]  Michael W. Mahoney Randomized Algorithms for Matrices and Data , 2011, Found. Trends Mach. Learn..

[12]  R. Schilling Measures, Integrals and Martingales , 2006 .

[13]  J. Cavanaugh,et al.  The Bayesian information criterion: background, derivation, and applications , 2012 .

[14]  A. V. D. Vaart,et al.  Asymptotic Statistics: Frontmatter , 1998 .

[15]  Min Yang,et al.  Information-Based Optimal Subdata Selection for Big Data Linear Regression , 2017, Journal of the American Statistical Association.

[16]  Michael Jackson,et al.  Optimal Design of Experiments , 1994 .

[17]  Kani Chen,et al.  Strong consistency of maximum quasi-likelihood estimators in generalized linear models with fixed and adaptive designs , 1999 .

[18]  Yun Yang,et al.  Communication-Efficient Distributed Statistical Inference , 2016, Journal of the American Statistical Association.

[19]  W. Newey,et al.  Large sample estimation and hypothesis testing , 1986 .

[20]  Yves G. Berger,et al.  Empirical likelihood confidence intervals for complex sampling designs , 2016 .

[21]  Shifeng Xiong,et al.  Some results on the convergence of conditional distributions , 2008 .

[22]  Carl-Erik Särndal,et al.  Model Assisted Survey Sampling , 1997 .

[23]  Rong Zhu,et al.  Optimal Subsampling for Large Sample Logistic Regression , 2017, Journal of the American Statistical Association.

[24]  Ruibin Xi,et al.  Aggregated estimating equation estimation , 2011 .

[25]  Dean P. Foster,et al.  New Subsampling Algorithms for Fast Least Squares Regression , 2013, NIPS.

[26]  A NeathAndrew,et al.  The Bayesian information criterion , 2012 .

[27]  Purnamrita Sarkar,et al.  A scalable bootstrap for massive data , 2011, 1112.5016.

[28]  F. Breidt,et al.  Local polynomial regresssion estimators in survey sampling , 2000 .

[29]  R. Kohn,et al.  Speeding Up MCMC by Efficient Data Subsampling , 2014, Journal of the American Statistical Association.

[30]  R. W. Wedderburn Quasi-likelihood functions, generalized linear models, and the Gauss-Newton method , 1974 .

[31]  Eric R. Ziegel,et al.  Multivariate Statistical Modelling Based on Generalized Linear Models , 2002, Technometrics.

[32]  Jie Tang,et al.  ArnetMiner: extraction and mining of academic social networks , 2008, KDD.