|
I'm writing the documentation for an open source machine learning library (for online Bayesian logistic regression), and would like to look at examples of truly excellent documentation that others have done. It's a Java library that's intended to be used in building applications, so I'm particularly interested in good examples of what to handle in Javadoc vs. the manual. We also have a standalone application for running experiments using the library, so pointers to exemplary documentation for standalone machine learning code are welcome as well. |
|
You can take a look at dlib. It's an open source C++ library I manage and I think it's documentation is quite good (I'm open to ideas for improvement though :) ). In my opinion, the two most helpful things are applying contract programming (e.g. DbC) to your ML code and then providing complete and detailed example programs. |
|
I can try and pitch our own project, scikit-learn, that is about machine learning too. I don't know if the documentation is excellent, but it is certainly a priority. Our approach is very much based on examples, with a large list of these, that are ran at documentation build time to generate the figures. |
