Abstract
In this research, a Bayesian hierarchical model was used to select and estimate variables.In the context of binary regression, existing approaches to variable selection in the context of binary classification are proposed. The proposed method is that the Laplace probability of the regression parameters is proposed and estimated with a Bayesian Markov chain Monte Carlo. The conceptual result is that by doing so, the regression model is transferred from a Gaussian framework to a full Laplacian framework without sacrificing With a lot of computational efficiency. In addition, the Gibbs sampler is effective for the Parameter estimation of the proposed model and is superior to the Metropolis algorithm Which has been used in previous studies on Bayesian binary regression. Both simulation studies and real data analysis indicate that the proposed method performs well compared to other binary regression methods.
