How do I pull a subset of an ARVIZ array within the inference data?

Questions v5
11

The only suggestion I’m having a hard time with is the constant_data used throughout this notebook.

Example:

hdi_helper = lambda ds: az.hdi(ds, input_core_dims=[["chain", "draw", "obs_id"]])
hdi_ppc = (
    pooled_trace.posterior_predictive.y.groupby(pooled_trace.constant_data.floor_idx)
    .apply(hdi_helper)
    .y
)
hdi_ppc

I found this in the docs…

InferenceData schema specification — ArviZ dev documentation

And while I see it in the inference objects shown in the ArviZ docs, I don’t see it in my inference object.