search

epinowcast / primarycensored

Primary event censored distributions in R and Stan.
https://primarycensored.epinowcast.org/
Other
7 stars 1 forks source link

primarycensoreddist -> primarycensored #108

Closed seabbs closed 4 days ago

seabbs commented 2 weeks ago

We now have a Julia version of this tool that is called PrimaryCensored to fit with the naming approach take in Distributions.jl does this mean we should consider renaming this to primarycensored (and so renaming all the functions that currently use dist as well or are we happy with these having slightly different names?

athowes commented 1 week ago

I prefer shorter version and think primarycensoreddist is a mouthful.

seabbs commented 1 week ago

This doesn't introduce a docs clash as the julia tooling is moving to PrimaryCensored.EpiAware.org

athowes commented 1 week ago

Regarding this sorry if this is something I've brought up before but I find the focus on "primary" censoring over censoring in general to be confusing.

I am using primarycensoreddist functions for double censored observation processes. Am I wrong to do so?

(I mention this as it relates to the name. Something like censordist seems cleaner and leave more options in future to expand into "not only primary censored".)

seabbs commented 1 week ago

its because the primary censoring is the special bit and the secondary censoring is the same as everything else. The point is that solving primary censoring is the challenge and the only new concept people need to worry about.

athowes commented 1 week ago

I think the exact technical details of the challenge matter more to developers than they do to users. Ultimately this solves both primary and secondary censoring. I expect most people will not be aware of this background.

seabbs commented 1 week ago

I don't agree. I think it adds an unnecessary layer of confusion and abstraction. I think we can move this to a discussion or new issue to expand the docs to better cover the gap you see.

It needs to be distinguished from normal censoring (i.e we can't use censoreddist) as that will just cause confusion. An double censoring is similarly a state of art term and so doesn't really help. Given that its important we clearly communicate what this actually does and in full generality that is primary censoring (i.e. we seek to support primary censored only use cases as well).

athowes commented 1 week ago

Sure. For example I think this documentation is overly focused on primary censoring:

This package provides both R functions for working with primary event censored distributions and Stan implementations for use in Bayesian modeling. Primary event censored distributions are useful for modeling delayed reporting scenarios in epidemiology and other fields. It provides support for arbitrary delay distributions, a range of common primary distributions, and allows for truncation and secondary event censoring to be accounted for.

Feel free to open an issue or discussion if you agree, otherwise happy to drop.

seabbs commented 1 week ago

I don't think I totally agree but happy to see a new issue. I do think the package overview needs a rework regardless to better capture use cases and expanded functionality

seabbs commented 1 week ago

I think the current conclusion is yes we should do this so moving from a question to issue. We need to: