Bayesian Hierarchical Modeling in Orbit

[brnjemison]

Hello,

I'm new to Bayesian stats and the Orbit library. Are there ways of creating hierarchical models in Orbit? For example, if I have a dataset similar to the iclaims dataset used in the Quick Start guide, however, an additional column for classification (i.e. state). I would like to use the knowledge from states with more extensive time series data to help the inform the posterior for states with less data. I've seen a handful of examples of this performed in PyMC3 and NumPyro and was wondering if there is similar functionality in Orbit. If so is there any documentation or tutorial where I can read up on this?
Thank you in advance.

[rjscout24]

you can create additional variables for your states, and then use the global estimates / guess as the prior along with appropriate distribution, as it has been shown also in pymc3 docs.