Documentation

[daniel-vainsencher]

In order of priority:

• For the UI, describe the parameters and what you do to them. To avoid giving a detailed explanation of the method, give a reference to an existing paper whose formulation (and preferably parameter names matches your own). For example, should I assume that lambda is the regularization strength? if so how is the default set? I have no idea what nlambda stands for. Etc.
• For implementations of published algorithms such as asdca, give references to paper describing what you implement. Add comments to the implementation that relate it to pseudo code there.

[mlcoding]

Thank you for your comments. These are what I will add to the "man" folder later as the documentation of the package, after the R and C codes are done.

[daniel-vainsencher]

It is easy to forget which code implements what algorithm, even for the implementer, never mind a reader; the best way to avoid this is to use informative names for functions as soon as you write them. Simple links to arxiv are also easy. More detailed documentation can certainly wait for later on.

[mlcoding]

Thank you. This is usually what I do and I`ll pay more attention to that.