Statistically and Computationally Efficient Estimating Equations for Large Spatial Datasets
Authors:
- Ying SunYSYing Sun
- Michael L. SteinMSMichael L. Stein
- 2 January 2016
- journal article
- Published by Taylor & Francis in Journal of Computational and Graphical Statistics
- Vol. 25 (1) , 187-208
Abstract
For Gaussian process models, likelihood based methods are often difficult to use with large irregularly spaced spatial datasets, because exact calculations of the likelihood for n observations require O(n3) operations and O(n2) memory. Various approximation methods have been developed to address the computational difficulties. In this paper, we propose new unbiased estimating equations based on score equation approximations that are both computationally and statistically efficient. We replace the inverse covariance matrix that appears in the score equations by a sparse matrix to approximate the quadratic forms, then set the resulting quadratic forms equal to their expected values to obtain unbiased estimating equations. The sparse matrix is constructed by a sparse inverse Cholesky approach to approximate the inverse covariance matrix. The statistical efficiency of the resulting unbiased estimating equations are evaluated both in theory and by numerical studies. Our methods are applied to nearly 90,000 satellite-based measurements of water vapor levels over a region in the Southeast Pacific OceanKeywords
This publication has 22 references indexed in Scilit:
- Parameter estimation in high dimensional Gaussian distributionsStatistics and Computing, 2012
- A Matrix-free Approach for Solving the Parametric Gaussian Process Maximum Likelihood ProblemSIAM Journal on Scientific Computing, 2012
- Covariance Tapering for Likelihood-Based Estimation in Large Spatial Data SetsJournal of the American Statistical Association, 2008
- Separable approximations of space‐time covariance matricesEnvironmetrics, 2007
- Approximate Likelihood for Large Irregularly Spaced Spatial DataJournal of the American Statistical Association, 2007
- Covariance Tapering for Interpolation of Large Spatial DatasetsJournal of Computational and Graphical Statistics, 2006
- Design and Modeling for Computer Experiments2005
- Matrix Preconditioning Techniques and Applications2005
- Models for Continuous Stationary Space-Time Processes1997
- A stochastic estimator of the trace of the influence matrix for laplacian smoothing splinesCommunications in Statistics - Simulation and Computation, 1990