Hi @OriolAbril,
Thank you for a great explanation! I see where I made the confusion, and the inclusion of the formulas for posterior and posterior predictive were quite helpful.
Just as follow up, you mentioned If you want to condition on different observations you need to pm.sample again to find the new posterior. Does that mean if I pass in new observed variables using the same model the sampling starts over as if there was no previous sampling, or the new observed variables are simply updating the previous values from the first sample?
Thank you again.