Plotting Posterior Predictive using arviz

Hey :slight_smile: in the tutorial here they use arviz.from_pymc3() in order to plot the Posterior predictive.
This method was depricated, and references to pymc3.to_inference_data. I can’t find online a useage of this function or it’s documentation.

Does anybody have an example? Thanks

1 Like

The pymc3 examples pages should be updated when there’s a new release. In the mean time you can check out the pymc-examples repo. I’ve got one here which does some posterior predictive plotting using up to date practice… pymc-examples/GLM-simpsons-paradox.ipynb at main · pymc-devs/pymc-examples · GitHub


It is also worth noting that to_inference_data is only available for PyMC3 4.x, which has not yet been released and doesn’t have docs available yet.

I think @drbenvincent linked notebook will already work because using return_inferencedata=True has already been updated to use the new conversion method under the hood, but you should still expect a bumpy ride when using PyMC3 4.x and you should see a warning about it being experimental for now at import time.