Machine Learning Toolbox
The machine-learning-toolbox repository makes it easy to create a Docker container for use during the Machine Learning Specialization.
Retrieve the machine-learning-toolbox build repository to create the container. Git, GraphLab, Jupyter, and Python are installed as part of the image.
$ git clone https://github.com/gdhorne/machine-learning-toolbox
For ease of instantiating an instance of the container image a script named 'container.sh' can be used to manage the entire lifecycle. For Microsoft Windows users it is recommended that Git Bash be installed instead of the standard Git software because it provides an *nix-like command line environment.
Create the container, optionally mapping a host file system share for storage. The file system share name '/home/me/datascience' is user selectable and host file system dependent. If no local file system share is desired simply omit the fourth argument '/home/me/datascience'. The container instance name 'toolbox', in these instructions, is user selectable at time of creation.
$ ./container.sh create toolbox gdhorne/machine-learning-toolbox /home/me/datascience
Apple Mac OS X: /Users/username/directory
Microsoft Windows: /c/Users/directory (allegedly)
Verify the container 'toolbox' has been successfully created and is running.
$ ./container.sh status
Stop the container 'toolbox'.
$ ./container.sh stop toolbox
Start the container 'toolbox'.
$ ./container.sh start toolbox
To learn more about the container lifecycle management features supported by 'container.sh' type,
$ ./container.sh --help
After creating the container these applications are accessible within a web browser.
Git: Accessible via WeTTY Jupyter: http://127.0.0.1:8888 Python: Accessible via WeTTY WeTTY: http://127.0.0.1:8000 UserID: mlt Password: science To enable the terminal/console management utility type 'screen' and press ENTER.
Alternatively, the machine-learning-toolbox image provides a traditional command line interface, without WeTTY, to some applications such as Git, Python, and vim. For convenience the terminal/console management utility 'screen' has been installed and starts automatically.
$ ./container.sh attach toolbox
Press ENTER if the container's shell prompt does not appear. To exit the container and leave it running press CTRL+P, CTRL+Q; this is the preferred method. To exit the container and stop it type 'exit'.
The GraphLab Create package available for Python requires either an academic licence or a commercial licence. Participants in the Unversity of Washington
Machine Learning Specialization delivered on the Coursera platform qualify for an academic licence subject to verification by Dato. Register and await an automated email containing a product licence key. Access the command line via WeTTY as indicated above. Installation of GraphLab Create can be initiated as shown.
$ ./bin/graphlab.sh EMAIL_ADDRESS LICENCE_KEY
Replace EMAIL_ADDRESS and LICENCE_KEY with the credentials sent to you via email.