Scalable MCMC for Large Data Problems Using Data Subsampling and the Difference Estimator
Sveriges Riksbank Working Paper Series No. 306
32 Pages Posted: 4 Jun 2016
Date Written: August 1, 2015
We propose a generic Markov Chain Monte Carlo (MCMC) algorithm to speed up computations for datasets with many observations. A key feature of our approach is the use of the highly effcient difference estimator from the survey sampling literature to estimate the log-likelihood accurately using only a small fraction of the data. Our algorithm improves on the O(n) complexity of regular MCMC by operating over local data clusters instead of the full sample when computing the likelihood. The likelihood estimate is used in a Pseudo-marginal framework to sample from a perturbed posterior which is within O(m -1/2) of the true posterior, where m is the subsample size. The method is applied to a logistic regression model to predict firm bankruptcy for a large data set. We document a significant speed up in comparison to the standard MCMC on the full dataset.
Keywords: Bayesian inference, Markov Chain Monte Carlo, Pseudo-marginal MCMC, estimated likelihood, GLM for large data
JEL Classification: C11, C13, C15, C55, C83
Suggested Citation: Suggested Citation