documentation

[tpapp]

I find this package extremely useful, and I think that it is sufficiently mature that it would benefit from documentation. I am opening this issue to discuss how to do this, and would be happy to contribute PRs.

What I am missing the most is details on computations, especially invariants that need to be satisfied, at least approximately. Eg

using WeightedOnlineStats
stat = WeightedMean()
fit!(stat, x, weights)
m = mean(stat)

then the expected invariant is

$$ m = \frac{\sum_i x_i w_i}{\sum_i w_i} $$

The example is deliberately trivial; for other moments I am not a 100% sure what the result is without looking at the code in detail. Formulas in docstrings would help. A lot of this is already used in tests.

It would also be great to have references to algorithms where applicable.

Setting up automatic docs generation using Documenter.jl would be a nice start, even if it is WIP. Ideally I would prefer to make small PRs.

[gdkrmr]

I have made some basic Documenter.jl work and there is a rudimentary documentation available now under https://www.guido-kraemer.com/WeightedOnlineStats.jl/

I'll leave this issue open to track the details for adding more documentation