SparseReg Toolbox for Matlab

SparseReg toolbox is a collection of Matlab functions for sparse regressions.

The toolbox is developed by Hua Zhou and Brian Gaines.

Compatibility

The code is tested on Matlab R2017a, but should work on other versions of Matlab with no or little changes. Current version works on these platforms: Windows 64-bit, Linux 64-bit, and Mac (Intel 64-bit). Type computer in Matlab’s command window to determine the platform.

Installation (Matlab version >= 2014b)

Download the Matlab toolbox installation file SparseReg.mltbx. Double click the downloaded file and you should be good to go. If it does not work for some reasons, follow the below instructions for Matlab version < 2014b.

Installation (Matlab version < 2014b)

1. Download ZIP File file using the links on the left.
2. Extract the zip file.

   unzip Hua-Zhou-SparseReg-xxxxxxx.zip
   

3. Rename the folder from Hua-Zhou-SparseReg-xxxxxxx to SparseReg.

   mv Hua-Zhou-SparseReg-xxxxxxx SparseReg
   

4. Add the SparseReg folder to Matlab search path. Start Matlab, cd to the SparseReg directory, and execute the following commands
   addpath(pwd) %<-- Add the toolbox to the Matlab path
save path %<-- Save for future Matlab sessions
5. Go through following tutorials for the usage. For help of individual functions, type ? followed by the function name in Matlab.
6. (Occasionally) you need to re-compile the Fortran code for your specific platform. First make sure the Matlab supported Fortran compiler is available on your system and type mex -setup FORTRAN to set up Matlab compilier utility. Within Matlab, enter the /private folder and type make to compile the Fortran source code.

Tutorial

• Sparse linear regression (enet, power, log, MC+, SCAD)
• Sparse generalized linear model (GLM) (enet, power, log, MC+, SCAD)
• Sparse generalized estimation equation (GEE) (enet, power, log, MC+, SCAD)

How to cite

If you use this toolbox in any way, please cite the software itself along with at least one publication or preprint.

• Software reference
  H Zhou and B Gaines. Matlab SparseReg Toolbox Version 1.0.0, Available online, March 2017.
• H Zhou, A Armagan, and D Dunson (2012) Path following and empirical Bayes model selection for sparse regressions. [arXiv:1201.3528]
• Default article to cite for least squares + generalized lasso penalty
  H Zhou and K Lange (2013) A path algorithm for constrained estimation, Journal of Computational and Graphical Statistics, 22(2):261-283.
• Default article to cite for convex loss + generalized lasso penalty
  H Zhou and Y Wu (2014) A generic path algorithm for regularized statistical estimation, Journal of American Statistical Association, 109(506):686-699.
• Default article to cite for path following in constrained convex programming
  H Zhou and K Lange (2015) Path following in the exact penalty method of convex programming, Computational Optimization and Applications, 61(3):609-634.
• Default article to cite for constrained lasso path following
  B Gaines and H Zhou (2018) Algorithms for Fitting the Constrained Lasso, Journal of Computational and Graphical Statistics, 27(4):861-871.

Contacts

Hua Zhou huazhou@ucla.edu

Brian Gaines brgaines@ncsu.edu