[WIP] fix hdi false positive warning

[OriolAbril]

Description

hdi used to support 2d data at most, so plot_hdi converted the data to 2d. Thanks to #1117 this is not necessary anymore and triggers this warning.

I have also done a couple of updates to summary also as a followup of #1117. I want to note that now az.hdi and xr.Dataset.quantile return an object with the same format and we can therefore use the same pattern as hdi now uses to report also equal tail intervals.

Checklist

• Follows official PR format
• New features are properly documented (with an example if appropriate)?
• Includes new or updated tests to cover the new feature
• Code style correct (follows pylint and black guidelines)
• Changes are listed in changelog

[codecov]

Codecov Report

Merging #1241 into master will increase coverage by 14.87%.
The diff coverage is 89.58%.

@@             Coverage Diff             @@
##           master    #1241       +/-   ##
===========================================
+ Coverage   78.27%   93.14%   +14.87%     
===========================================
  Files          98       98               
  Lines        9726     9721        -5     
===========================================
+ Hits         7613     9055     +1442     
+ Misses       2113      666     -1447     

Impacted Files	Coverage Δ
arviz/plots/plot_utils.py	94.57% <ø> (ø)
arviz/plots/hdiplot.py	89.23% <83.87%> (+5.30%)	⬆️
arviz/plots/backends/bokeh/hdiplot.py	100.00% <100.00%> (ø)
arviz/plots/backends/matplotlib/hdiplot.py	92.30% <100.00%> (+7.69%)	⬆️
arviz/stats/stats.py	96.24% <100.00%> (+0.03%)	⬆️
arviz/plots/pairplot.py	92.77% <0.00%> (ø)
arviz/utils.py	89.56% <0.00%> (+0.35%)	⬆️
arviz/data/inference_data.py	80.11% <0.00%> (+1.15%)	⬆️
arviz/data/base.py	100.00% <0.00%> (+10.97%)	⬆️
arviz/data/io_cmdstan.py	94.84% <0.00%> (+51.07%)	⬆️
... and 8 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 32be6fc...c1af041. Read the comment docs.

[AlexAndorra]

Thanks for fixing this! I had it on my to do list for weeks now and never found the time 🤦‍♂️ Will review tomorrow 👌 Le dim. 14 juin 2020 à 00:47, codecov[bot] <[email protected]> a écrit :

…

Codecov <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=h1> Report Merging #1241 <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=desc> into master <https://codecov.io/gh/arviz-devs/arviz/commit/31aac4c1084e40b99da0024218431a800ba222c9&el=desc> will *increase* coverage by 0.00%. The diff coverage is 87.50%. [image: Impacted file tree graph] <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=tree> @@ Coverage Diff @@ ## master #1241 +/- ## ======================================= Coverage 93.16% 93.16% ======================================= Files 98 98 Lines 9714 9715 +1 ======================================= + Hits 9050 9051 +1 Misses 664 664 Impacted Files <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=tree> Coverage Δ arviz/plots/hdiplot.py <https://codecov.io/gh/arviz-devs/arviz/pull/1241/diff?src=pr&el=tree#diff-YXJ2aXovcGxvdHMvaGRpcGxvdC5weQ==> 83.01% <0.00%> (-0.91%) ⬇️ arviz/stats/stats.py <https://codecov.io/gh/arviz-devs/arviz/pull/1241/diff?src=pr&el=tree#diff-YXJ2aXovc3RhdHMvc3RhdHMucHk=> 96.23% <100.00%> (+0.02%) ⬆️ ------------------------------ Continue to review full report at Codecov <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=continue>. *Legend* - Click here to learn more <https://docs.codecov.io/docs/codecov-delta> Δ = absolute <relative> (impact), ø = not affected, ? = missing data Powered by Codecov <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=footer>. Last update 31aac4c...032429b <https://codecov.io/gh/arviz-devs/arviz/pull/1241?src=pr&el=lastupdated>. Read the comment docs <https://docs.codecov.io/docs/pull-request-comments> . — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1241 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHIJMTG67JF42TGWWVQQRY3RWP6YVANCNFSM4N5FMMZQ> .

[AlexAndorra]

Looks nice! Just added some comments 😉

[AlexAndorra]

All looks good now, thanks @OriolAbril 👌

[OriolAbril]

Ready to merge, if nobody complains I'll merge tomorrow

[OriolAbril]

I guess there are a couple points that could be a bit more problematic:

• I am setting the coordinate values in hdi dim to ["lower", "higher"] which does not include the hdi_prob info (this diverges from summary df for instance, I find hdi_3% a little confusing). I have included hdi_prob as an attribute in case it were needed.
• passing 2d data to hdi will work and be backwards compatible but I have added a FutureWarning because it does not comply with ArviZ chain, draw, *shape convention. In a future release we should remove FutureWarning and enforce the convention to make hdi behave coherently with all other ArviZ functions

Extra easter egg: now az.hdi's behaviour is very similar to the point that using hdi_data allows to pass the result of Dataset.quantile to az.plot_hdi (a transpose may be needed to make sure the quantile/hdi dimension is the last one)

[AlexAndorra]

That looks good to me 👌 Can be included in 0.8.4