Matthieu Gomez

Is not it what dplyr does already (ie skip missing in filter but not in mutate)? On Mon, Oct 19, 2020 at 1:13 PM Milan Bouchet-Valat < [email protected]> wrote: >...

Ok, we can use CSV.write

@salbert83 Does your quantile function gives the same thing as Base Julia for weights equal to one?

There are several ways to extend unweighted quantile for non-frequency weights. The current implementation satisfies these three properties: 1. equals unweighted quantile for a weight vector of ones 2. gives...

In the pdf, you don't show that 1, 2, 3 are true, you just look at one example.

Right, 2&3 are satisfied. And I agree with you — I don't particularly like the definition of unweighted quantile used in Julia. But I still think it's important to match...

I guess you could open a PR to add a `type` kwarg to `quantile` in Statistics, adding the type you want + open a PR in StatsBase to add a...

Oh, I see! So yeah, it'd be terrific to add the kwargs `alpha` and `beta` to the weighted quantile function in `StatsBase` (at least if you find an extension that...

I believe it should work for any AbstractVector.

> As KrylovKit tries to assume as little as possible about the type of vectors that the user works with, No but, come on, in this instance, KrylovKit *assumes* that...