Online Program Home
My Program

Abstract Details

Activity Number: 248
Type: Contributed
Date/Time: Monday, August 1, 2016 : 2:00 PM to 3:50 PM
Sponsor: Section on Statistical Learning and Data Science
Abstract #320511 View Presentation
Title: Bayesian Regression Using a Prior on the Model Fit
Author(s): Brian Naughton* and Howard Bondell
Companies: North Carolina State University and North Carolina State University
Keywords: Bayesian ; Sparse Regression ; High-dimensional ; Shrinkage priors

Analyzing data in a Bayesian framework has the advantage of including subjective prior knowledge in the model. Examples include imposing realistic constraints on the parameters, including scientific knowledge, or incorporating results from previous studies such as in a meta-analysis. However, implementing a subjective Bayesian approach in practice proves difficult, particularly when modeling multiple parameters as in regression. We propose a Bayesian linear model by placing a prior directly on the coefficient of determination instead of the regression coefficients, allowing researchers to easily include information through a univariate parameter. We also propose using the model as a shrinkage prior for regularization, handling sparsity and high-dimensional modeling.

Authors who are presenting talks have a * after their name.

Back to the full JSM 2016 program

Copyright © American Statistical Association