March 26, 2023, 12:03am
I came across the Bambi library and have been playing with it using the
Using a relatively simple mode:
bmb.Model("Reaction ~ 0 + Days + (1 | Subject)", sleep), how might I recreate it using base pymc? The bambi model produces the following graph
which, I can’t seem to figure out how to reproduce in base pymc. In particular, I’m confused how the deterministic block functions in this case.
The deterministic block is implementing the “non-centered parameterization” of a hierarchical prior. There’s a nice discussion of this in the
radon example notebook. McElreath also has a detailed discussion of why you want to do this in this video.
This is the source code of a talk I gave showing how to do the same model with both PyMC and Bambi
Let me know if that helps
Edit: This may also help
unfinished-ideas/sparse.ipynb at main · bambinos/unfinished-ideas · GitHub
Notice the goal there was to replace some indexing operation with a sparse dot product. But there you have how to reproduce the model with PyMC