Docker based setup for scikit http://scikit-learn.org/stable/index.html and mxnet https://mxnet.apache.org/

Windows

1. Install docker: https://docs.docker.com/docker-for-windows/install/
2. Share the drive where these files are located: https://docs.docker.com/docker-for-windows/#docker-settings
3. Build the image: docker build -t scikit .
4. Run the example plotter: docker run -v %cd%:/mnt/scikit --rm -i -t scikit python /mnt/scikit/plot_classifier_comparison.py
5. Check the output: output\plot_classifier_comparison.png
6. Run a my_script.py in a different directory: docker run -v /path/to/script:/mnt/scikit --rm -i -t scikit python /mnt/scikit/my_script.py

OSX

1. Install docker: brew install docker
2. Build the image: docker build -t scikit .
3. Run the example plotter: docker run -v $PWD:/mnt/scikit --rm -i -t scikit python /mnt/scikit/plot_classifier_comparison.py
4. Check the output: output\plot_classifier_comparison.png
5. Run a my_script.py in a different directory: docker run -v /path/to/script:/mnt/scikit --rm -i -t scikit python /mnt/scikit/my_script.py

Notes

Since there is no window manager, you will likely need to modify your script to render figures to an image file, ex:

import matplotlib
matplotlib.use('Agg')
# Do things
figure.savefig('/mnt/scikit/plot_classifier_comparison.png')