update Moderation example to best practice and v4

[drbenvincent]

Updates the moderation analysis example (see #170) to current best practice.

Because this is just a minor update with no substantive changes, I propose to speed things up that we don't necessarily need 2 reviewers. The only meaningful change was to shift all custom plot functions into a single (hidden) code cell.

Note: in pm.sample, I'm currently having to provide nuts={"target_accept": 0.9} rather than just target_accept=0.9 (see pymc-devs/pymc#5620)

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[OriolAbril]

I think this is breaking the docs build. Can you try removing the year key altogether?

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

OriolAbril commented on 2022-03-20T12:52:26Z
----------------------------------------------------------------

PyMC3 -> PyMC

drbenvincent commented on 2022-03-20T13:35:04Z
----------------------------------------------------------------

done

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

OriolAbril commented on 2022-03-20T12:52:27Z
----------------------------------------------------------------

PyMC3 -> PyMC

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

OriolAbril commented on 2022-03-20T12:52:27Z
----------------------------------------------------------------

Line #2.        COORDS = {"obs": np.arange(len(x))}

There is no need to define coords with aranges in v4 provided they are first use (and can therefore be inferred automatically) in a Data object.

drbenvincent commented on 2022-03-20T13:35:29Z
----------------------------------------------------------------

have removed coords

[drbenvincent]

done

---

View entire conversation on ReviewNB

[drbenvincent]

have removed coords

---

View entire conversation on ReviewNB

[drbenvincent]

It's not adding the pm.* tags.

But docs now seems to work :)

[drbenvincent]

Everything looks good, but still no pm.* tags

[OriolAbril]

Everything looks good, but still no pm.* tags

I think we still need to update the style guide, but you can see that CI is passing without them. They should not be present anymore (and should be removed from the ones that have them).

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

OriolAbril commented on 2022-03-20T16:29:34Z
----------------------------------------------------------------

I mentioned the explanation category in the myst view because its where I noticed, but I should probably done that here. if both files are updated without syncing them then only the changes in the most recent one are kept :/

The last paragraph about mediation analysis should also have a link to the mediation notebook (and viceversa I guess, not sure if the mediation has been updated already)

drbenvincent commented on 2022-03-20T17:14:09Z
----------------------------------------------------------------

Have attempted to cross reference the two notebooks

[drbenvincent]

Have attempted to cross reference the two notebooks

---

View entire conversation on ReviewNB

[drbenvincent]

Notebook cross references now work fine

[OriolAbril]

Thanks! How has the extra myst step in pre-commit worked out for you?

[drbenvincent]

For these kinds of small changes I don't think there was much difference. But I think it will definitely be more useful with larger changes. Better diff.

I was having some issues when running the pre commit on a target notebook:

I ended up running pre-commit run --all

[OriolAbril]

I have never tried that on specific files and the error doesn't make much sense to me, could it be because it was unstaged (before doing git add that then shows in git status in green)?