How to keep dimensions in a posterior predictive sample?

Questions v5
3

Hi,

Thanks! I think you’re right that I’ve got dimensions wrong, in some way.

What I’d like to do is get separate samples of Event Counts for both Green and Blue ‘Event Types’. And a plot_ppc that plots Green and Blue Events separately.

Does that make sense?

The ‘posterior’ estimated by sampling is giving me potential ‘mu’ values, and it gives me different mu values for different Event Types, which is what I want -

image
image1075×233 28.7 KB

How do I get posterior predictive samples and plots split up in the same way, by ‘Event Type’?

Thank you!