Abstract:Bayesian models are generally computed with Markov Chain Monte Carlo (MCMC) methods. The main disadvantage about MCMC methods is the large number of iterations they need to sample the posterior distributions of model parameters, especially for large data sets. On the other hand, variable selection remains a challenging problem due to its combinatorial search space, where Bayesian models are a promising solution. In this work, we study how to accelerate Bayesian model computation for variable selection in linear regression. We propose a fast Gibbs sampler algorithm, a widely used MCMC method, that incorporates several optimizations. We use non-informative and conjugate prior distributions on several model parameters, which enable data set summarization in one pass exploiting an augmented set of sufficient statistics. Thereafter the algorithm can iterate in main memory. Sufficient statistics are indexed with a sparse binary vector to efficiently compute matrix projections based on selected variables. Discovered variable subsets probabilities, selecting and discarding each variable, are stored on a hash table for fast retrieval in future iterations. We study how to integrate our algorithm into a database management system (DBMS), exploiting aggregate User-Defined Functions for parallel data summarization and stored procedures to manipulate matrices with arrays. An experimental evaluation with real data sets evaluates accuracy and time performance, comparing our DBMS-based algorithm, with the R package. Our algorithm is shown to produce accurate results, scale linearly on data set size and run orders of magnitude faster than the R package.

Robust Bayesian nonparametric variable selection for linear regression

Bayesian variable selection in linear regression models with instrumental variables

Bayesian variable selection in linear regression models with non-normal errors

Robust variable selection for partially linear additive models

Bayesian Variable Selection in Linear Regression in One Pass for Large Data Sets

Dynamic sparsity on dynamic regression models

Robust Bayesian model selection for heavy-tailed linear regression using finite mixtures

The spike-and-slab lasso and scalable algorithm to accommodate multinomial outcomes in variable selection problems

Bayesian variable selection for matrix autoregressive models

Scalable Bayesian bi-level variable selection in generalized linear models

Bayesian Variable Selection for Single Index Logistic Model

Bayesian Variable Selection and Estimation for Group Lasso

Fast Bayesian inference in a class of sparse linear mixed effects models

Robust Bayesian Model Averaging for Linear Regression Models With Heavy-Tailed Errors

Bayesian high-dimensional covariate selection in non-linear mixed-effects models using the SAEM algorithm

Penalized robust estimators in logistic regression with applications to sparse models

Application of Bayesian variable selection in logistic regression model

A novel Bayesian method for variable selection and estimation in binary quantile regression

Disjunct Support Spike and Slab Priors for Variable Selection in Regression under Quasi-sparseness

Scalable Bayesian variable selection and model averaging under block orthogonal design