Misspecified var_names

[ColCarroll]

In certain cases, if an element of var_names is not in the dataset, should we should throw an exception? Right not, it will plot any var_names it can find, and ignore the others. This behavior has actually been kind of nice when plotting a centered eight trace vs a non-centered eight trace, but confusing if you typo one of the var_names.

[ahartikainen]

Could it throw warning? Sometimes there are "missing" parameters.

Also can do some word2vec similarity scoring what was probably meant? This could be helpful for complex names?

[ColCarroll]

Have been convinced that this is ok.

[canyon289]

I have same sentiment as @ahartikainen Should we at least print something to the console? Right now behavior is confusing if you don't remember the exact var_name.

[ColCarroll]

right now the behavior is equivalent to data.posterior[var_names] (and specifically, data.posterior[var_names].sel(**coords), though, which I like: i.e., matching xarray's behavior.

[canyon289]

Fair point, matching xarrays behavior is good idea! We can focus this discussion on how verbose we want to be to user.

[michaelraczycki]

Is it still relevant? I'd like to add this check and warnings if it's still needed

[OriolAbril]

Try someting like:

idata = az.load_arviz_data("rugby")
az.plot_posterior(idata, var_names=("atts", "unexisting_var"))

and see what happens