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.

Application of Bayesian variable selection in logistic regression model

A Novel Variational Bayesian Method for Variable Selection in Logistic Regression Models.

Bayesian variable selection in linear regression models with instrumental variables

Bayesian Variable Selection for Single Index Logistic Model

Fully Bayesian logistic regression with hyper-LASSO priors for high-dimensional feature selection

Study of Bayesian variable selection method on mixed linear regression models

Variable Selection in Logistic Regression Model with Genetic Algorithm.

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

Multivariate Bayesian variable selection with application to multi-trait genetic fine mapping

Comprehensive Stepwise Selection for Logistic Regression

Simultaneous Dimension Reduction and Variable Selection for Multinomial Logistic Regression

Robust Bayesian nonparametric variable selection for linear regression

Variable selection for partially linear models via Bayesian subset modeling with diffusing prior

Robust adaptive LASSO in high-dimensional logistic regression

Bayesian Fused Lasso Modeling for Binary Data

High-dimensional Feature Selection Using Hierarchical Bayesian Logistic Regression with Heavy-tailed Priors

A robust and efficient estimation and variable selection method for partially linear models with large-dimensional covariates

Scalable Bayesian variable selection for structured high‐dimensional data

Variable Selection Using the Lasso-Cox Model with Bayesian Regularization

Scalable Bayesian bi-level variable selection in generalized linear models

Bayesian Variable Selection with Related Predictors