A0384
Title: Inference for spatial autoregressive models using stochastic gradient descent
Authors: Ji Meng Loh - New Jersey Institute of Technology (United States) [presenting]
Gan Luan - New Jersey Institute of Technology (United States)
Abstract: The stochastic gradient descent (SGD) procedure is considered to fit spatial auto-regressive models to lattice data, incorporating a recently developed perturbation method to obtain standard errors in addition to model parameter estimates. We derive the SGD update equations and will present the results of a simulation study to examine the empirical coverage of confidence intervals constructed using the perturbation procedure.
