usability of StatsBase documentation

[alanedelman]

Happened to be looking to do, what i now know is sample( Weights( [1,2,1] )) . I searched the doc for nonuniform, and 'non uniform'. Nothing showed up. Then tried non-uniformand found sample but searching the page didn't really have that term.

I finally found what i wanted, but didn't know the word Weights so had to poke around some more. A couple of examples would have been very helpful.

Here's towards making julia stats more usable....

[andreasnoack]

Here's towards making julia stats more usable....

Did you intend to add more?

It would be useful with a more detailed description of what would be required here. That will make it more likely to be fixed. Adding more examples to the documentation more generally is a broader issue that will require more labor than is currently available for maintenance of this repo.

[alanedelman]

Suppose we donated labor -- e.g. the julialab might have a docuthon creating examples in some orderly fashion?
I gather you would want examples and doctest? How can we make this work in a way that would make the StatsBase organization happy?

[bkamins]

My experience with DataFrames.jl is that it is more efficient that several tutorials NOT within StatsBase.jl repo were written and e.g. blogged about.

Then such tutorials (if they positively pass the quality check) can be given a cross-linked in the documentation of StatsBase.jl as an additional material or included in the official documentation (potentially after some mutation).

The benefit of this approach is that it will have a much quicker turnover than going through PR review process and is easier to experiment with (which eventually can help us find the best way of presenting the available functionality).

Regarding Weights I think that the major the problem is that the documentation there is too advanced for a novice. But this is exactly the role that the tutorials could serve (most novice users are not aware that you can have different types of weights).