We should support a few basic machine learning algorithms from the start. These include

• naive bayes
• random forests of decision trees (ensemble learning)
• neural networks
• support vector machines (kernel learning)
• others as folks are interested

We should have a working version of each of these ready to go so that we could repeat the Biswas et al. (2013) study trivially if desired.

More MLA are available and developers can add them as they like, but we should consider support for these 3 as fundamental requirements of the pipeline.

Each of these needs to be documented within the Sphinx pages for the document. That should include references to the literature.