Note: the codes in this repository are deprecated and superseded by the more sophisticated package projpred, which has many more features and is easier to use.
This repository contains simple example codes for performing projection predictive variable selection for linear regression models using Stan+R. The method is described in detail in the document. See also our comparative study for performance evaluation of the method (Piironen and Vehtari, 2016).
You should have RStan installed. Run first
library(rstan)
source('projection.R')
After that you can run one of the main scripts, either
source('search.R')
or
source('search_cv.R')
Both of these will take some time to execute, but you can reduce the running time by reducing the number of samples generated by Stan (see files search.R and search.R)
Piironen, J. and Vehtari, A. (2016). Comparison of Bayesian predictive methods for model selection. Statistics and Computing (online)
This software is distributed under the GNU General Public Licence (version 3 or later); please refer to the file licence.txt, included with the software, for details.