Removed repetitive variable names from forest plots of multivariate variables #1527
Conversation
AlexAndorra
added
Enhancement
There was a problem hiding this comment.
This looks great, thanks @fonnesbeck ! I don't think we need this for more plots -- or maybe plot_parallel ? Curious what the others think.
The test failure is weird, looks like a problem with Azure
|
How does this look in the case were there are no verbose coords? |
|
@ahartikainen if the variable is not a multivariate random variable, then it uses the variable name. The only change should be in the multivariate case. |
|
@AlexAndorra yes, they look like resource provisioning errors. |
|
Actually, I noticed a bug with multiple non-combined chains:
Don't merge until I fix this. |
|
Restarting azure and now it seems to run without problem |
Codecov Report
@@ Coverage Diff @@
## main #1527 +/- ##
==========================================
+ Coverage 91.77% 91.79% +0.01%
==========================================
Files 105 105
Lines 11330 11342 +12
==========================================
+ Hits 10398 10411 +13
+ Misses 932 931 -1
Continue to review full report at Codecov.
|
|
OK, fixed. A multi-trace multivariate forest plot now looks like this:
whereas a forest plot of three univariate variables looks like this:
|
…ariables (arviz-devs#1527) * Removed repetitive variable names in forest plots for mutivariate variables * Removed errant indents * Applied Black formatting * Added changes to CHANGELOG * Fixed multi-chain labelling bug
Forest plots of multivariate variables currently prepend the variable name to each element label, which adds a lot of clutter to the plot. This removes the variable names from all but the first element.
Can add this feature to other plots as needed, either here or in a separate PR.
Not sure what sort of test would be relevant here, if any.
Closes #1524