uncover ML

Machine learning tools for the Geoscience Australia uncover project.

Quickstart

Before you start, make sure your system has the following packages installed,

• gdal (libgdal-dev)
• openmpi
• hdf5

We strongly recommend using a virtual environment. To install, simply run setup.py:

$ python setup.py install

or install with pip:

$ pip install git+https://github.com/GeoscienceAustralia/uncover-ml.git@release

The python requirements should automatically be built and installed.

Cubist

In order to use the cubist regressor, you need to first make sure cubist is installed. This is easy with our simple installation script, invoke it with:

$ ./makecubist <installation-path>

Once cubist is installed, it will add a configuration file to the script, if you like, you can test that it's been installed in the correct place by checking the contents of uncover-ml/cubist_config.py, its presence indicates that the installation completed successfully.

Next you need to rerun the setup script with:

$ python setup.py install

Which will ensure the cubist_config has been added successfully. Now you should be able to use the cubist regressor in the pipeline file.

Running

See the usage documentation.

Running on NCI

Please see The PBS Readme .

Collaboration

This software is jointly developed by NICTA and Geoscience Australia. For a list of features still to be implemented, see the issue tracker.

Useful Links

Home Page

http://github.com/GeoscienceAustralia/uncover-ml

Documentation

http://GeoscienceAustralia.github.io/uncover-ml

Issue tracking

https://github.com/GeoscienceAustralia/uncover-ml/issues

Bugs & Feedback

For bugs, questions and discussions, please use Github Issues.