Moritz Schauer

Hm, with `nu = 0`, this is not a proper distribution and one cannot sample anymore from it, compare https://github.com/JuliaStats/ConjugatePriors.jl/blob/8e3a6a79c7f1c9d9efd5720506219929c089933a/src/normalgamma.jl#L48 The originating package `Distribution.jl` in general only supports proper distributions....

Does it make sense to discuss this in the larger context of the package? Are our other implementations typically "as fast as reasonably can be" or is there a continuous...

Yes, we should expose that. A possible function name would be `medians` giving a range or a pair, or is this too subtile?

As said elsewhere, it would be nice if Julia would develop a standard design pattern for this. In accumulation where there is no sensible typestable default value, you end up...

Not sure if it helps you but DNF gave me me a quirky snipped which initialises an accumulator in the loop in a way Julia can handle ``` function myfsum(f,...

`countmap`, the name is not so lucky, the `counts(x, k::Integer, [wv::AbstractWeights])` ("If an integer k is provided, only values in the range 1:k will be considered.") is a bit non-sensical,...

> counts(Dict/Vector, x). That reads right. > Applying a binning map is what histograms do, right? The actual work in histograms is to find the binning map, right? Otherwise it...

One more thing > `rle` and `inverse_rle` are not really related to statistics. I think the keyword is "average run length", which is important for statistics and the reason why...

It doesn’t help that the ˋcorrected = trueˋ version of ˋcovˋ which is applied multiplies everything with ˋNaN == 1/(n-1)ˋ

Unitful defines `isnan` https://github.com/PainterQubits/Unitful.jl/blob/2be7e0692cafd948b66132920ee407b73220717f/src/quantities.jl and base should perhaps define it for `TimePeriod`s as well.