edgarst/dogss: Sparse-Group Bayesian Feature Selection with Expectation Propagation
Given a response vector y and data matrix X as well as a grouping of features G, this recovers a vector of coefficients beta with y = X*beta + error, where beta is sparse on a between- and within-group level. The algorithm to recover the parameters is called expectation propagation and is much faster than Gibbs sampling.
Getting started
Browse package contents
Package details |
|
|---|---|
| Author | Edgar Steiger |
| Maintainer | Edgar Steiger <edgar.steiger@gmail.com> |
| License | think-about-license... |
| Version | 1.0 |
| URL | https://github.com/edgarst/dogss |
| Package repository | View on GitHub |
| Installation |
Install the latest version of this package by entering the following in R:
|
