Allow Unknown DNC in Community Level

[BrettBoval]

Situation

The state of Iowa has decided as of 1 April to no longer provide postive case reports at the state or county level (see Iowa Department of Health and Human Services Press Release). Florida has suspended reporting "due to a technical issue" and the odds don't look good that they will restart.

Complication

Divergent from CDC

We deliberately had a different implementation than the CDC, which more prominently flagged places where we believed the case data should be treated with caution.

The current implementation from libs/metricslcommunity_levels.py#L20:

   # TODO(michael): The CDC footnotes say:
    #     If the number of cases in 7 days for a jurisdiction is missing, the
    #     7-day case rate is assigned to the “low” category. If both 7-day
    #     admissions and 7-day percentage inpatient beds indicators are N/A, the
    #     community burden category is assigned N/A.
    #
    # For now I'm allowing 1 hospital metric to be missing, but not allowing
    # cases to be missing since that is rare and usually indicates we're
    # blocking data or something. In that case, I'd rather have no community
    # level calculated. But we can revisit if it ends up being a problem.

Implementation Artifact in our Filtering

Our current implementation of forward filling heuristics + don't-grade-stale-data creates unexpected behaviour to the end user.

Specifically, we usually ingest cumulative data and then calculate the diffs to create a daily new timeseries. In the case where the cumulative number is the same day to day, it can be difficult to tell the difference between "the county did not report today" and "the county reported affirmatively that there were no new cases". On a scraper by scraper basis there may be some meta-text or context to discern between the two, but we currently don't have a full solution for this.

Our current solution has two heuristics that are temporally staggered (by somewhat chance) by one day.

• We treat trailing zeros (a.k.a. all the most recent data is zeros) with suspicion. For the first x days of consecutive zeros, we use the last non-zero value as the latest value for above-the-fold metrics. so [15,0,0,0,0,0,0] returns 15 as the "latest" value.
• Eventually, after x days of consecutive zeros, we give up waiting, and let all those zeros propagate turning Daily New Cases (DNC) to zero. So [15,0,...,0,0,] returns 0 once the gap is long enough.
• Separately, we have a different process that looks at all our datastreams, and masks (likely too aggressively) any timeseries where the day is more than y days stale. This turns the metric to Unknown, which bubbles up and will turn a metric gray.

The complication is that x and y are both ~14 days, but one of them is calculated before applying a np.diff which shifts it by one day. So for the first ~13 days (I'm being deliberately vague on defining a day here) we return "last non-zero value in series" and then on one day the "if y days stale mask everything" returns Unknown, and then the following day the case timeseries gets stale enough that we let the zeros forward fill and it returns a zero.

The exact number of zeros to be confirmed.

• t-1 day:[15,0,0,0,0,0,0,0,0,0,0,0,0,0] pipes out [15] 13 days ago which passes the stale test which then counts as 15 for risk level calculation
• t-0 day:[15,0,0,0,0,0,0,0,0,0,0,0,0,0,0] pipes out [15] 14 days ago which fails the stale test which then returns None for risk level calculation
• t+1 day:[15,0,0,0,0,0,0,0,0,0,0,0,0,0,0,0] pipes out [0] 1 day ago which passes the stale test which then returns 0 for risk level calculation

This causes the map and the location page to transiently filter to gray and then green.

Resolutions

Summary

1. Update community risk level logic to handle unknown DNC as 0 (the lesser of two evils)
2. Add the appropriate disclaimers to Iowa and Florida in our normal disclaimer code.
3. Tweak the "turn this unknown after x days to something longer than the DNC ffill code.
4. Explore special casing the DNC where we freeze it as Unknown starting from a specific day which overrides the "turn it unknown and hide it".

	Current	Proposed	Ideal
US Map Color	Grey/Green	Green	Green w/ Flag
Community Risk Level	Grey/Green	Green	Green w/ Flag
DNC Latest Value	None/0	None/0	None
Timeseries	Masked/ffilled	Masked/ffilled	Frozen @ Last Reporting

[smcclure17]

Action item from this PR #1368