New CRAN release?

[n-kall]

There have been several small features included since the last CRAN release, including:

• thin_draws now does automatic thinning
• subset_draws has exclude option (currently if a user uses exclude = TRUE with CRAN version, it just ignores it silently, leading to the opposite behaviour as expected)
• pareto_smooth can be used properly with log weights (in weight_draws, but also the are_log_weights argument)
• new user facing functions pareto_* diagnostics
• new extract_variable_array function

Soon there should be larger changes merged (e.g. weighted draws updates, changes to summarise_draws backend to use rvars) which might constitute a larger release. But as the currently merged features are now being used in public case studies (e.g. from @avehtari), perhaps it is a good time to push a point update to CRAN? @paul-buerkner @mjskay

[mjskay]

+1

I am not teaching this quarter, so I will also start to have time to iterate on the bigger things again too. I like the idea of a small release first.

[avehtari]

• It would be nice to include PR export helper functions #361, if we can agree on the names
• There are some issues that might be small enough to fix?
  • subset_draws duplicates variables if variables argument contains NA #365
  • pareto_khat should always return the same data structure #359
  • export ps_khat_threshold #358
  • In summarise draws, a function that returns a character changes all columns to character #355
  • pareto diagnostics return named values, preventing renaming #346
  • Add functions for PIT and LOO-PIT #338
  • Vignette for Pareto diagnostics #322
  • pareto_khat errors when it should just give a warning #315
  • piping weight_draws / resample_draws #310

EDIT: formatting

[paul-buerkner]

I don't really have time to work much on posterior at the moment. Any of these issues that @mjskay, @n-kall or you can address?

[n-kall]

Any of these issues that @mjskay, @n-kall or you can address?

Yes, I should be able to prepare PRs for at least those related to pareto diags. A couple of these issues will need decisions regarding naming e.g. #361, #346, which might need further input from others

[mjskay]

I can probably take a look at some of the others in about a week and a half (at a conference currently)

[paul-buerkner]

Thank you both!

[n-kall]

Several of the bugs @avehtari listed have now been fixed. @mjskay Any thoughts on the others / should they wait until the next CRAN release?

[paul-buerkner]

I unfortunately don't have time to work more on posterior at the moment. But I am happy to do a new CRAN release now. @avehtari what are your thoughts?

[avehtari]

I'd like to include the new Pareto-k-diagnostic vignette #372

[avehtari]

The Pareto vignette has been merged. I also don't have time now to add or fix anything else, so ok with the new CRAN release

[paul-buerkner]

Okay. Will submit it today

[mjskay]

Thanks - sorry I didn’t get to other issues, it’s been a crazy quarter. I’m hoping to swing back around on some of my outstanding issues this summer, including bigger changes like weighted rvars, but those can wait until after this next release.

[n-kall]

I see 1.6.0 now on CRAN, thanks @paul-buerkner ! Also, should the pkgdown website be rebuilt or is it an automatic process?

[paul-buerkner]

I will have to do this. Will do today. Am Do., 4. Juli 2024 um 07:40 Uhr schrieb Noa Kallioinen < ***@***.***>:

…

I see 1.6.0 now on CRAN, thanks @paul-buerkner <https://github.com/paul-buerkner> ! Also, should the pkgdown website be rebuilt or is it an automatic process? — Reply to this email directly, view it on GitHub <#367 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ADCW2ADQJDCKUFVTC7EBPTLZKTN57AVCNFSM6AAAAABHQEQFNKVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDEMBYGE2TMNJTGM> . You are receiving this because you were mentioned.Message ID: ***@***.***>