Skip to main content

Picasso Python Package

Project description

PICASSO: Penalized Generalized Linear Model Solver - Unleash the Power of Non-convex Penalty

Unleash the power of nonconvex penalty

L1 penalized regression (LASSO) is great for feature selection. However when you use LASSO in very noisy setting, especially when some columns in your data have strong colinearity, LASSO tends to give biased estimator due to the penalty term. As demonstrated in the example below, the lowest estimation error among all the lambdas computed is as high as 16.41%.

Requirements

  • Linux or MacOS

It may take lots of effort to build on Windows. One way to do it is using CMAKE and MSVC. Be careful of issues like the system bits.

Installation

In the following process, you may need to be root (sudo).

Install from source file (Github):

  • Clone picasso.git via git clone https://github.com/jasonge27/picasso.git

  • Make sure you have setuptools

    Using Makefile

  • Run sudo make Pyinstall command.

    Using CMAKE

  • Build the source file first via the cmake with CMakeLists.txt in the root directory. (You will see a .so or .lib file under (root)/lib/ )

  • Run cd python-package; sudo python setup.py install command.

Install from PyPI:

  • pip install pycasso
  • Note: Owing to the setting on different OS, our distribution might not be working in your environment (especially in Windows). Thus please build from source.

You can test if the package has been successfully installed by:

import pycasso
pycasso.test()

Usage

import pycasso
x = [[1,2,3,4,5,0],[3,4,1,7,0,1],[5,6,2,1,4,0]]
y = [3.1,6.9,11.3]
s = pycasso.Solver(x,y)
s.train()
s.predict()

For Developer

Please follow the sphinx syntax style

To update the document: cd doc; make html

Copy Right

Author:Jason Ge, Haoming Jiang
Maintainer:Haoming Jiang <jianghm@gatech.edu>

Project details


Release history Release notifications

This version
History Node

0.1.1

History Node

0.1.0

History Node

0.0.6.dev2

History Node

0.0.5.dev2

History Node

0.0.5.dev0

History Node

0.0.3.dev0

Download files

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

Filename, size & hash SHA256 hash help File type Python version Upload date
pycasso-0.1.1.tar.gz (1.2 MB) Copy SHA256 hash SHA256 Source None Jan 1, 2018

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging CloudAMQP CloudAMQP RabbitMQ AWS AWS Cloud computing Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page