Quidel covidtest tests_per_device

[krivard]

PR for tests_per_device

[krivard]

Try a correlation within a region across time, as recommended by Roni in the Quidel channel.

[jingjtang]

Needs to be solved

************* Module delphi_quidel_covidtest.run
delphi_quidel_covidtest/run.py:26:0: R0912: Too many branches (16/12) (too-many-branches)
delphi_quidel_covidtest/run.py:26:0: R0915: Too many statements (58/50) (too-many-statements)

Resolved --20200822

[jingjtang]

Rank Correlation across time at state and msa level:

Revised(2020-08-21):

• Compare Quidel covid_ag_smoothed_test_per_device with jhu-csse confirmed_7dav_incidence_prop
• negative correlations still exist
• add number of MSA with correlations larger than 0
  QuidelCovidCorrelationAnalysis.pdf

[krivard]

The negative correlations are particularly suspicious for including TX. Meet with Jeremy and Roni to figure out whether we should publish this signal.

[krivard]

• Figure out whether to comment out tests per device or disable/exclude it using configuration
• Fix git history to drop the cache files and resolve conflicts with main

[jingjtang]

GeoMapping related work after talking to Dmitry:

• refactor the geo_map.py to make the steps for msa/hrr_to_parent_state independent from traditional geo mapping
• jingjing will talk with Roni next Friday about whether to stick with geographical pooling (considering parent states).

[krivard]

@jingjtang Do those tasks need to be completed before this can be merged? If so we should convert to draft.

[jingjtang]

@jingjtang Do those tasks need to be completed before this can be merged? If so we should convert to draft.

Nope. We could solve them in another PR.

[krivard]

Okay, tracking those next steps in #253 then.

For this PR, it remains to:

• rebase to drop ght commits
• pass review

[krivard]

Linter passes, tests pass on a fresh checkout. Technically I authored this so I can't review it, but Jingjing did all the work to bring it up to snuff so... :pulls lever: