Skip to content
glmnet for python
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Glmnet for python


Han Fang


Using pip (recommended)

pip install glmnet_py

Complied from source

git clone
cd glmnet_py
python install

Requirement: Python3, Linux

Currently, the checked-in version of is compiled for the following config:

Linux: Linux version 2.6.32-573.26.1.el6.x86_64 (gcc version 4.4.7 20120313 (Red Hat 4.4.7-16) (GCC) ) OS: CentOS 6.7 (Final) Hardware: 8-core Intel(R) Core(TM) i7-2630QM gfortran: version 4.4.7 20120313 (Red Hat 4.4.7-17) (GCC)


import glmnet_py
from glmnet import glmnet

For more examples, see


This is a python version of the popular glmnet library (beta release). Glmnet fits the entire lasso or elastic-net regularization path for linear regression, logistic and multinomial regression models, poisson regression and the cox model.

The underlying fortran codes are the same as the R version, and uses a cyclical path-wise coordinate descent algorithm as described in the papers linked below.

Currently, glmnet library methods for gaussian, multi-variate gaussian, binomial, multinomial, poisson and cox models are implemented for both normal and sparse matrices.

Additionally, cross-validation is also implemented for gaussian, multivariate gaussian, binomial, multinomial and poisson models. CV for cox models is yet to be implemented.

CV can be done in both serial and parallel manner. Parallellization is done using multiprocessing and joblib libraries.

During installation, the fortran code is compiled in the local machine using gfortran, and is called by the python code.

+Getting started:

The best starting point to use this library is to start with the Jupyter notebooks in the test directory (glmnet_examples.ipynb). Detailed explanations of function calls and parameter values along with plenty of examples are provided there to get you started.


Algorithm was designed by Jerome Friedman, Trevor Hastie and Rob Tibshirani. Fortran code was written by Jerome Friedman. R wrapper (from which the MATLAB wrapper was adapted) was written by Trevor Hastie.

The original MATLAB wrapper was written by Hui Jiang (14 Jul 2009), and was updated and is maintained by Junyang Qian (30 Aug 2013).

This python wrapper (which was adapted from the MATLAB and R wrappers) was originally written by B. J. Balakumar (5 Sep 2016), later modified by Han Fang.

B. J. Balakumar, (5 Sep 2016). Department of Statistics, Stanford University, Stanford, California, USA.


  • Friedman, J., Hastie, T. and Tibshirani, R. (2008) Regularization Paths for Generalized Linear Models via Coordinate Descent, Journal of Statistical Software, Vol. 33(1), 1-22 Feb 2010

  • Simon, N., Friedman, J., Hastie, T., Tibshirani, R. (2011) Regularization Paths for Cox's Proportional Hazards Model via Coordinate Descent, Journal of Statistical Software, Vol. 39(5) 1-13

  • Tibshirani, Robert., Bien, J., Friedman, J.,Hastie, T.,Simon, N.,Taylor, J. and Tibshirani, Ryan. (2010) Strong Rules for Discarding Predictors in Lasso-type Problems, Stanford Statistics Technical Report

You can’t perform that action at this time.