prepare for v2.6.0

[jgabry]

Branch for preparing the 2.6.0 release. @avehtari I will wait to merge this and do the release until the vignette in #210 is ready so that we can include it (I think it's almost ready).

@LeeviLindgren This branch also includes a few minor changes to the loo_predictive_metric and loo_[s]crps functions that will be included in this release:

• All of these functions now have a cores argument passed to psis
• loo_[s]crps : change ll argument to log_lik to match loo_predictive_metric

I think eventually we should also add a psis_object argument to these functions so that it's possible to avoid rerunning psis internally (see #213) but we can do that in the future.

[codecov-commenter]

Codecov Report

Merging #214 (ea33ab7) into master (1560636) will increase coverage by 0.21%.
The diff coverage is 100.00%.

❗ Current head ea33ab7 differs from pull request most recent head 1623fb2. Consider uploading reports for the commit 1623fb2 to get more accurate results

@@            Coverage Diff             @@
##           master     #214      +/-   ##
==========================================
+ Coverage   93.09%   93.31%   +0.21%     
==========================================
  Files          30       30              
  Lines        2768     2768              
==========================================
+ Hits         2577     2583       +6     
+ Misses        191      185       -6     

Impacted Files	Coverage Δ
R/crps.R	100.00% <100.00%> (+10.34%)	⬆️
R/loo_predictive_metric.R	100.00% <100.00%> (ø)

📣 We’re building smart automated test selection to slash your CI/CD build times. Learn more

[jgabry]

@avehtari I think this release is ready. I'm about to be traveling for a few days, but I will submit to CRAN next week when I'm back.

Draft of release notes:

loo v2.6.0

• New loo_predictive_metric() function for computing estimates of leave-one-out
  predictive metrics: mean absolute error, mean squared error and root mean
  squared error for continuous predictions, and accuracy and balanced accuracy for
  binary classification. (Additional loo utilities #202, @LeeviLindgren)

• New functions crps(), scrps(), loo_crps(), and loo_scrps() for
  computing the (scaled) continuously ranked probability score. (CRPS and SCRPS #203, @LeeviLindgren)

• New vignette "Mixture IS leave-one-out cross-validation for high-dimensional Bayesian models." This is a demonstration of the mixture estimators proposed in Silva and Zanella (2022).

[avehtari]

proposed in Silva and Zanella (2022). -> proposed by Silva and Zanella (2022).

[jgabry]

@avehtari I'm going to submit this release to CRAN and we can include #216 in the next release (we're still waiting on a few people). CRAN wants a release by April 1 (to fix a test that fails in an upcoming version of R) and I'm going to be pretty busy the next few days so I will submit it now.

[jgabry]

loo v2.6.0 is now on CRAN. I'm working on updating the loo website now

[avehtari]

Excellent!