Closed gvegayon closed 2 weeks ago
Attention: Patch coverage is 95.83333%
with 3 lines
in your changes missing coverage. Please review.
Project coverage is 94.73%. Comparing base (
3d35d13
) to head (7b5063e
). Report is 4 commits behind head on main.
:umbrella: View full report in Codecov by Sentry.
:loudspeaker: Have feedback on the report? Share it here.
This is now ready for review, @damonbayer.
At some point, files under docs/source/tutorials
were removed unintentionally. 1a75196 fixes that. Also, a figure for testing was added outside of model/
. 9b2d667 fixes the latter.
It's a bit odd that all of the hospitalization code lives in the latent
sub-package, because it handles observations as well. Also, the weekday effect is definitely more related to observations. Also, I'm not sure what the process
sub-package is supposed to be for. Seems like it has some generic math and some things specific to Rt. Not sure what to make of these observations. Perhaps related to https://github.com/CDCgov/multisignal-epi-inference/issues/163.
It's a bit odd that all of the hospitalization code lives in the
latent
sub-package, because it handles observations as well. Also, the weekday effect is definitely more related to observations. Also, I'm not sure what theprocess
sub-package is supposed to be for. Seems like it has some generic math and some things specific to Rt. Not sure what to make of these observations. Perhaps related to #163.
That is too broad of a comment for this PR :). Perhaps submit a few tickets? A couple of answers:
1 and 3 are helpful comments. Thanks.
I don't quite follow 2, but I think I may be misled by the variable names in this PR https://github.com/CDCgov/multisignal-epi-inference/blob/a6cfc9568b038f7ef6f1c97ca715aa469f085f2a/model/src/pyrenew/latent/hospitaladmissions.py#L192-L195
The observed_hosp_admissions
actually get passed later to an observation process as a mean parameter, right?
1 and 3 are helpful comments. Thanks.
I don't quite follow 2, but I think I may be misled by the variable names in this PR
The
observed_hosp_admissions
actually get passed later to an observation process as a mean parameter, right?
100% agree the names are misleading. I've updated them across the functions, tests, and tutorial. It should be ready for a final review now.
arrayutils.PeriodicBroadcaster
.process.RtPeriodicDiffProcessSample
to leverage thePeriodicBroadcaster
class.PeriodicEffect
andWeeklyEffect
, the former usingPeriodicBroadcaster
.PeriodicBroadcaster
works + a case sampling from a transform Dirichlet distribution (for the Day of the week effect).