Skip to main content

A regression solver for linear and quantile regression models and lasso based penalizations

Project description

asgl package


asgl is a Python package that solves several regression related models for simultaneous variable selection and prediction, in low and high dimensional frameworks. This package is directly related to research work shown on this paper.

The current version of the package supports:

  • Linear regression models
  • Quantile regression models

And considers the following penalizations for variable selection:

  • No penalized models
  • lasso
  • group lasso
  • sparse group lasso
  • adaptive lasso
  • adaptive group lassso
  • adaptive sparse group lasso


The package makes use of some basic functions from scikit-learn and numpy, and is built on top of the wonderful cvxpy convex optimization module. It is higly encouraged to install cvxpy prior of the installation of asgl following the instructions from the original authors, that can be found here). Additionally, asgl makes use of python multiprocessing module, allowing, if requested, for parallel execution of the code highly reducing computation time.

Usage example:

In the following example we will analyze the BostonHousing dataset (available here). Even though the asgl package can easily deal with much more complex datasets, we will work using this one so we are not affected by computation time. We will show how to implement cross validation on a grid of possible parameter values for an sparse group lasso linear model, how to find the optimal parameter values and finally, how to compute the test error.


The following code performs cross validation in a grid of different parameter values for an sparse group lasso model on the well known BostonHousing dataset:

# Import required packages
import numpy as np
from sklearn.datasets import load_boston
import asgl

# Import test data #
boston = load_boston()
x =
y =
group_index = np.array([1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5])

# Define parameters grid
lambda1 = (10.0 ** np.arange(-3, 1.51, 0.2)) # 23 possible values for lambda
alpha = np.arange(0, 1, 0.05) # 20 possible values for alpha

# Define model parameters
model = 'lm'  # linear model
penalization = 'sgl'  # sparse group lasso penalization
parallel = True  # Code executed in parallel
error_type = 'MSE'  # Error measuremente considered. MSE stands for Mean Squared Error.

# Define a cross validation object
cv_class = asgl.CV(model=model, penalization=penalization, lambda1=lambda1, alpha=alpha,
                   nfolds=5, error_type=error_type, parallel=parallel, random_state=99)

# Compute error using k-fold cross validation
error = cv_class.cross_validation(x=x, y=y, group_index=group_index)

num_models, k_folds = error.shape
# error is a matrix of shape (number_of_models, k_folds)
print(f'We are considering a grid of {num_models} models, optimized based on {k_folds}-folds cross validation')

# Obtain the mean error across different folds
error = np.mean(error, axis=1)

For a full review on the capabilities of these package we suggest accessing the user_guide notebook provided in the GitHub repository. Additionally, you can find more here and here.


If you use ASGL for academic work, we encourage you to cite our paper. Thank you for your support and we hope you find this package useful!

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

asgl-1.0.5.tar.gz (16.9 kB view hashes)

Uploaded source

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page