Cosmetic fixes to getting started notebook.

[rpgoldman]

There were some problems with the version on the PyMC3 docs site.
Fixed some warnings. Tidied up some markup (some of which didn't seem to process properly on gh pages), and added a note explaining about pm.Deterministic

[review-notebook-app]

Check out this pull request on 

Review Jupyter notebook visual diffs & provide feedback on notebooks.

---

Powered by ReviewNB

[codecov]

Codecov Report

Merging #4062 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #4062   +/-   ##
=======================================
  Coverage   86.83%   86.83%           
=======================================
  Files          88       88           
  Lines       14165    14165           
=======================================
  Hits        12300    12300           
  Misses       1865     1865           

[AlexAndorra]

LGTM, thanks @rpgoldman ! I had just one small suggestion about Deterministic

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

AlexAndorra commented on 2020-08-21T08:38:55Z
----------------------------------------------------------------

Maybe show how to do it here, as it's very short?rate = pm.Deterministic("rate", pm.math.switch(switchpoint >= years, early_rate, late_rate))

rpgoldman commented on 2020-08-21T14:41:52Z
----------------------------------------------------------------

You are quite right. I have put that in, and will push it to the MR momentarily. Thanks for having a look!

[rpgoldman]

You are quite right. I have put that in, and will push it to the MR momentarily. Thanks for having a look!

---

View entire conversation on ReviewNB

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

fonnesbeck commented on 2020-08-21T14:50:14Z
----------------------------------------------------------------

similarly here, no context required

rpgoldman commented on 2020-08-21T14:54:34Z
----------------------------------------------------------------

It's actually necessary here now. The reason is that az.summary() internally calls az.from_pymc3() to build an InferenceData, and that function needs information from the model in order to do the translation. If you leave it out, you get a warning.

If you look at the version on the PyMC3 docs page, https://docs.pymc.io/notebooks/getting_started.html you will see the warning.

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

fonnesbeck commented on 2020-08-21T14:50:14Z
----------------------------------------------------------------

Not a big deal, but the model context is not necessary here.

rpgoldman commented on 2020-08-21T14:55:39Z
----------------------------------------------------------------

See my response above: ArviZ now needs the model for from_pymc3() (this is necessary to appropriately allocate random variables among the various groups in the InferenceData.

[rpgoldman]

It's actually necessary here now. The reason is that az.summary() internally calls az.from_pymc3() to build an InferenceData, and that function needs information from the model in order to do the translation. If you leave it out, you get a warning.

If you look at the version on the PyMC3 docs page, https://docs.pymc.io/notebooks/getting_started.html you will see the warning.

---

View entire conversation on ReviewNB

[rpgoldman]

See my response above: ArviZ now needs the model for from_pymc3() (this is necessary to appropriately allocate random variables among the various groups in the InferenceData.

---

View entire conversation on ReviewNB

[fonnesbeck]

OK, cool. Did not realize that.

[fonnesbeck]

LGTM then.

[rpgoldman]

OK, cool. Did not realize that.

Yes, originally ArviZ was using a sneaky pointer to the model that was stashed in an _strace, which made everyone very nervous, and wouldn't work when loading a saved trace.