ENH: use numpyro implementation of dismod at to derive a consistent remission rate #1
Conversation
aflaxman
requested review from
albrja,
hussain-jafari,
patricktnast,
rmudambi and
stevebachmeier
as code owners
see 2024_10_30b_test_dismod_for_consistent_artifact_data.ipynb for example
see 2024_10_30b_test_dismod_for_consistent_artifact_data.ipynb for details
aflaxman
changed the title
| def dismod_f(t, y, args): | ||
| S, C = y | ||
| i, r, f, m = args | ||
| return (-m * S - i * S + r * C, -m * C + i * S - r * C - f * C) |
There was a problem hiding this comment.
I object to the approach isort/black has taken to formatting this
There was a problem hiding this comment.
This is probably controversial because I know people are used to writing differential equations in a certain way, but if this is going to be code maybe you should make each item in this tuple a variable, and give everything here meaningful names instead of single letters?
| if rate_name in art.keys: | ||
| art.replace(rate_name, df_out) | ||
| else: | ||
| art.write(rate_name, df_out) |
There was a problem hiding this comment.
Is there a reason you put this in the builder.py and not loader.py?
There was a problem hiding this comment.
Is there a reason you put this in the builder.py and not loader.py?
based on this slack discussion, it seemed most convenient (since it is available in make_artifacts). How should I think about the difference between builder and loader?
tweak model specs to test this more
Title: Use a custom implementation of the DisMod-AT approach to get remission, since it is not available from get_draws
Description
Changes and notes
This refactors a bunch of statistical computation code from my notebooks into a .py file and eventually will use it.Verification and Testing
I'm not sure how to store multiple keys in the artifact which I think I should do, since the code here will find different draws from incidence, prevalence, etc that are consistent with the draws of remission it generates.