Eight schools noncentered model in PyMC

[fonnesbeck]

Updated eight schools noncentered model, using the current version of PyMC.

[MansMeg]

Great!

Two things.

1. You now committed the json data file. We keep the data as zip, so it must have been an accident?
2. Also, there should not be any point of having an additional draws object from pyMC? It should be a true reference draw already.

So I suggest you remove these parts.

[fonnesbeck]

Ah, forgot to zip them -- thanks!

[fonnesbeck]

@MansMeg we are having a posteriordb hackathon today, so there should be a few PRs rolling in over the next few hours/days. Happy to lend a hand if you need.

Let me know if anything else needs changing for this example.

[fonnesbeck]

Is there anything else required here?

[MansMeg]

Sorry. Courses are starting so its currently a lot pn my table.

[fonnesbeck]

Understood, no problem! We (the PyMC team) are hoping to implement a bunch of these models in the current version of PyMC over the next few weeks or so. Given your constraints, would you recommend having us fork the project and then you can move them over when you have time?

[MansMeg]

Hi! Yes. That is probably the best if you want to use the models asap.

[JTorgander]

Thank you for contributing and sorry for the delay! I have made a minor adjustment to the prior to have it match our Stan model. I have also removed the pymc reference draws since our goal is to use the same draws between PPL's as much as we can.

[fonnesbeck]

Terrific. Anything you need me to do prior to merging?

[JTorgander]

LGTM. @MansMeg?

[MansMeg]

LGTM - feel free to merge Jakob and then we can soon start to do a new release of posteriordb.