Docker image for reproducible machine learning in Python3

Have you ever been bothered by building machine learning environments? This docker image is for you! Let's enjoy machine learning in the reproducible enviromnent easily.

Features

• Ensure reproducibility by full version spesification
• Based on Anaconda (4.2.0) enviroments

Installed Packages

• numpy 1.11.1
• scipy 0.18.1
• scikit-learn 0.17.1
• pandas 0.18.1
• matplotlib 1.5.3
• and so on.

See package_list.txt for details.

Usage

$ docker pull sotetsuk/py3-machine-learning

Run interactively

$ docker run -it sotetsuk/py3-machine-learning

Run Jupyter Notebook

$ docker -d -p 8888:8888 sotetsuk/py3-machine-learning /run_jupyter.sh

with sharing current directory:

docker run -v $(pwd):/work -d -p 8888:8888 sotetsuk/py3-machine-learning /run_jupyter.sh

Run as IntelliJ (PyCharm) interpreter

1. Install Python and Docker integration plugins.
2. Follow these steps:
   • [Create New Project]
   • [Python]
   • Python SDK [NEW]
   • [Add Remote]
   • Finally,
   • Check [Docker]
   • Server: [New] => [OK] (change nothing)
   • Image name: sotetsuk/py3-machine-learning:latest (or another version)
   • Python interpreter path: python3

Versioning

• MAJOR: Some imcompatible change
• MINOR: Update versions of some packages
• PATCH: Add other packages

How to contribute

Any contribution is more than welcome! Fork the repo and send PR!

LICENSE

MIT